Examining expression of immune genes in devils with DFTD

Source codes: https://github.com/markziemann/dftd_immune_expression

Background

We have designed a target enrichment panel to examine gene expression of immune genes. This is so we can inexpensively profile immune gene expression in devils. We will be looking at a lot of different variables such as dftd-free vs affected, DFTD1 vs DFTD2, etc.

The main comparisons are:

-Wild vs Captive devils

-Healthy vs DFTD infected devils

-DFT1 and DFT2 devils

I have their sex/age/seasons categories there as well if we decide to go with more detail analysis in the future, but I think just stick with these three for now. Some devils are a part of the Captive control and Wild control groups . There are a couple samples that won’t go into the analysis that I made note of. Meliza and Simba.

suppressPackageStartupMessages({
    library("reshape2")
    library("gplots")
    library("DESeq2")
    library("mitch")
    library("limma")
    library("kableExtra")
    library("dplyr")
    library("edgeR")
    library("eulerr")
    library("UpSetR")
})

Functions.

maplot <- function(de,contrast_name) {
  sig <-subset(de, padj < 0.05 )
  up <-rownames(subset(de, padj < 0.05 & log2FoldChange > 0))
  dn <-rownames(subset(de, padj < 0.05 & log2FoldChange < 0))
  GENESUP <- length(up)
  GENESDN <- length(dn)
  DET=nrow(de)
  SUBHEADER = paste(GENESUP, "up, ", GENESDN, "down", DET, "detected")
  ns <-subset(de, padj > 0.05 )
  plot(log2(de$baseMean),de$log2FoldChange, 
       xlab="log2 basemean", ylab="log2 foldchange",
       pch=19, cex=1, col="dark gray",
       main=contrast_name, cex.main=0.7)
  points(log2(sig$baseMean),sig$log2FoldChange,
         pch=19, cex=1, col="red")
  mtext(SUBHEADER,cex = 0.7)
  grid()
}

make_volcano <- function(de,name) {
    sig <- subset(de,padj<0.05)
    N_SIG=nrow(sig)
    N_UP=nrow(subset(sig,log2FoldChange>0))
    N_DN=nrow(subset(sig,log2FoldChange<0))
    DET=nrow(de)
    HEADER=paste(N_SIG,"@5%FDR,", N_UP, "up", N_DN, "dn", DET, "detected")
    plot(de$log2FoldChange,-log10(de$padj),cex=1,pch=19,col="darkgray",
        main=name, xlab="log2 FC", ylab="-log10 pval", xlim=c(-6,6))
    mtext(HEADER)
    grid()
    points(sig$log2FoldChange,-log10(sig$padj),cex=1,pch=19,col="red")
}

Load data

Here we load the data in from the aligner.

tmp <- read.table("3col.tsv.gz")
x <- as.data.frame(acast(tmp, V2~V1, value.var="V3"))
colnames(x) <- gsub("fastq/","",colnames(x))
dim(x)

## [1] 310 143

Remove poor samples

Samples with <1M reads should be omitted. Will also round values to integers.

cs <- colSums(x)
cs <- cs[order(cs)]

barplot(cs,ylim=c(1e2,2e7),log="y",main="All samples",cex.names=0.5)

barplot(head(cs),log="y",main="Samples with low reads",cex.names=0.7)

abline(h=1e6,col="red",lty=2)

x <- x[,colSums(x)>1e6]
x <- round(x)

Identify poorly represented genes

rc <- rowMeans(x)
rc <- rc[order(rc)]
barplot(rc+1,log="y")

hist(log(rc+1))

round(rc[which(rc<10)],digits=1)

##   CNTF   CTF1   PTX3   RAG2    OSM   AIF1  IL17B IL36RN  STAT2 CLEC4E   CD3g 
##    0.0    0.0    0.0    0.0    0.0    0.2    0.4    0.4    0.5    0.5    0.6 
##   CSF2 CXCL13   IL25  Casp9 CX3CL1 CCLD16  IL17A  IFNB1   CSF3  CCL19  CCL26 
##    0.7    0.8    0.8    0.9    0.9    1.3    1.5    1.8    2.0    2.3    2.4 
##  IFNA1  FCAMR  KITLG  IL17F CCLD15    LIF  CCL22   IL11  IFNL1   IL33   IL26 
##    2.5    2.5    2.7    3.2    3.4    3.9    3.9    4.0    4.4    4.5    4.6 
##  IL12A IL1F10  RPS18   CD70  CCL27  ACKR4 IL36L1    FGB IL31Ra CXCL11 CXCL14 
##    4.9    5.0    5.0    5.3    5.6    5.8    6.0    6.2    6.8    7.5    7.7 
## IL36L2 CXCL17    IL4    IL9  CCL17  CCLD4 
##    8.1    8.1    8.2    8.3    8.6    9.8

A lot of the shorter genes (cytokines) had lower expression.

Sample sheet

ss <- read.table("SampleSheet.txt",sep="\t",header=TRUE)

ss %>%  kbl(caption="Sample sheet")%>% kable_styling("hover",full_width=FALSE) #Makes the sample sheet table 

Sample sheet

Lane	Project	Client.ID	Sample	X	Category	Sex.Age.Season	Sex	Age	Season	Location	Agebyyear	Weight	X.1	Number.of.Tumours	Ulcerated	UlcerationDepth	Secondary.infection	Total.tumour.volume	Barcode.sequence	PF.Clusters	X..of.the.lane	X..Perfect.barcode	X..One.mismatch.barcode	Yield..Mbases.	X..PF.Cluster	X…..Q30	Mean.Quality	X.2	X.3	X.4
1	default	S1_LauraGrace	6102461	NA	Healthy/Wild	FAP	F	A	Spring	Channel	2.6219178	7.2	NA	NA	NA		NA	NA	AGAGTAGC+AAGGCGTA	4,511,702	1.00	98.29	1.71	1,363	100	93.82	35.92	NA	NA	NA
1	default	S1_Kaoota	6102462	NA	N/A	FAA	F	A	Autumn	Channel	4.1808219	10	NA	3	TRUE	superficial	FALSE	279.276	ACGGAACA+ACGAGAAC	5,202,254	1.00	97.76	2.24	1,571	100	93.72	35.91	NA	NA	NA
1	default	S1_Sarafina	6102463	NA	DFT1	FAP	F	A	Spring	Channel	2.6191781	6.4	NA	NA	NA		NA	95140.790	ATCCAGAG+TAACGTCG	6,491,488	1.00	74.07	25.93	1,960	100	93.74	35.92	NA	NA	NA
1	default	S1_Lucy	6102464	NA	DFT1	FAU	F	A	Summer	Channel	2.9178082	7.1	NA	3	TRUE	deep	TRUE	700825.170	AAGCACTG+AGGTCAAC	6,121,785	1.00	98.00	2.00	1,849	100	93.84	35.93	NA	NA	NA
1	default	S1_Turmeric	6102465	NA	Healthy/Wild	FAU	F	A	Summer	Channel	1.9095890	7.8	NA	NA	NA		NA	NA	GAAGGAAG+GACGTCAT	6,479,047	1.00	97.28	2.72	1,957	100	93.85	35.94	NA	NA	NA
1	default	S1_Singapore	6102466	NA	Healthy	FAA	F	A	Autumn	Channel	5.0000000	8.6	NA	NA	NA		NA	NA	ACAGCTCA+TACTGCTC	4,481,213	1.00	97.99	2.01	1,353	100	93.95	35.96	NA	NA	NA
1	default	S2_Belinda	6102467	NA	Healthy/Wild	FAU	F	A	Summer	Channel	1.8657534	6.6	NA	NA	NA		NA	NA	GCAAGATC+AGTCGAAG	3,296,547	1.00	97.40	2.60	996	100	94.12	35.98	NA	NA	NA
1	default	S2_B99008	6102468	NA	Captive	FAU	F	A	Summer	Monarto	3.7369863	8.12	NA	NA	NA		NA	NA	GCTATCCT+AACAGGTG	5,338,890	1.00	98.08	1.92	1,612	100	93.69	35.91	NA	NA	NA
1	default	S2_Bonney	6102469	NA	Captive	FAU	F	A	Summer	Healesville	2.5863014	5.4	NA	NA	NA		NA	NA	ACGTTCAG+AGTTGTGC	14,968,459	3.00	97.82	2.18	4,520	100	93.68	35.91	NA	NA	NA
1	default	S2_Rosie	6102470	NA	Healthy/Wild	FAU	F	A	Summer	Channel	1.8904110	7.4	NA	NA	NA		NA	NA	CGCATGAT+AAGCCTGA	4,556,781	1.00	95.70	4.30	1,376	100	93.94	35.94	NA	NA	NA
1	default	S2_Hope	6102471	NA	DFT1	FAU	F	A	Summer	Channel	2.9315068	6.4	NA	4	TRUE	deep	TRUE	103815.530	AGCACTTC+ACGTCCAA	1,971,776	0.00	97.96	2.04	595	100	93.62	35.88	NA	NA	NA
1	default	S2_Sally	6102472	NA	DFT1	FAU	F	A	Summer	Channel	2.9945205	7	NA	1	TRUE	deep	FALSE	1588.272	CGTTGCAA+ATAGAGCG	2,429,449	1.00	97.42	2.58	734	100	93.78	35.93	NA	NA	NA
1	default	S3_Apu	6102473	NA	DFT1	MAA	M	A	Autumn	Channel	3.1150685	9.5	NA	1	FALSE	superficial	TRUE	200.000	CCTTGATC+GATGGAGT	4,939,580	1.00	96.94	3.06	1,492	100	93.71	35.91	NA	NA	NA
1	default	S3_Simone	6102474	NA	DFT1	FAW	F	A	Winter	Channel	3.2739726	10.2	NA	1	TRUE	superficial	TRUE	2670.678	CGGTTGTT+GTGGTATG	4,015,027	1.00	95.91	4.09	1,213	100	93.42	35.86	NA	NA	NA
1	default	S3_ST62	6102475	NA	Healthy/Wild	FAW	F	A	Winter	Channel	3.2712329	10.2	NA	NA	NA		NA	NA	TGAGCTAG+GAACGGTT	4,579,681	1.00	96.67	3.33	1,383	100	93.75	35.92	NA	NA	NA
1	default	S3_Jessie	6102476	NA	Captive	FAW	F	A	Winter	Monarto	2.1945205	12.95	NA	NA	NA		NA	NA	CCGTAAGA+AGAGTCCA	4,340,498	1.00	97.95	2.05	1,311	100	93.54	35.88	NA	NA	NA
1	default	S3_McCauley	6102477	NA	Captive	MAA	M	A	Autumn	Healesville	3.0328767	8.52	NA	NA	NA		NA	NA	CAACACCT+AGCTACCA	16,944,152	3.00	97.98	2.02	5,117	100	93.05	35.78	NA	NA	NA
1	default	S4_Kerchak	6102478	NA	Healthy	MAA	M	A	Autumn	Channel	4.1863014	11.6	NA	NA	NA		NA	NA	CGGCTAAT+CTCGTTCT	6,886,511	1.00	97.80	2.20	2,080	100	93.61	35.89	NA	NA	NA
1	default	S4_Duffman	6102479	NA	Healthy	MAA	M	A	Autumn	Channel	3.1123288	12.4	NA	NA	NA		NA	NA	GACGATCT+TCGTGCAT	8,301,336	2.00	97.75	2.25	2,507	100	93.83	35.93	NA	NA	NA
1	default	S4_Wiggum	6102480	NA	DFT1	MAA	M	A	Autumn	Channel	3.1178082	8.3	NA	2	FALSE		FALSE	7272.309	GTGAAGTG+GAGCAATC	4,201,355	1.00	97.17	2.83	1,269	100	93.64	35.90	NA	NA	NA
1	default	S4_Hogan	6102481	NA	Healthy	MAA	M	A	Autumn	Channel	3.1616438	11.6	NA	NA	NA		NA	NA	TACGCCTT+GCTACTCT	2,990,345	1.00	97.77	2.23	903	100	93.71	35.90	NA	NA	NA
1	default	S4_Pooh	6102482	NA	DFT1	MAA	M	A	Autumn	Channel	4.1863014	11.8	NA	1	FALSE		FALSE	7577.856	CTTCGACT+GATCTTGC	3,821,151	1.00	97.31	2.69	1,154	100	93.60	35.89	NA	NA	NA
1	default	S4_Milhouse	6102483	NA	DFT1	MAA	M	A	Autumn	Channel	4.1013699	11	NA	1	TRUE	deep	TRUE	480.000	AACCGTTC+CTAGCTCA	10,492,900	2.00	98.13	1.87	3,169	100	93.80	35.92	NA	NA	NA
1	default	S5_Fenton	6102484	NA	Captive	MAU	M	A	Summer	Monarto	3.0000000	N/A	NA	NA	NA		NA	NA	GATTACCG+TTGGACTG	5,125,566	1.00	96.21	3.79	1,548	100	93.89	35.94	NA	NA	NA
1	default	S5_Bryan Adams	6102485	NA	Captive	MAW	M	A	Winter	Healesville	4.1616438	8.6	NA	NA	NA		NA	NA	CACTAGCT+TCGGATTC	5,317,767	1.00	97.33	2.67	1,606	100	93.24	35.83	NA	NA	NA
1	default	S5_Silky	6102486	NA	Healthy/Wild	MAW	M	A	Winter	Channel	2.2849315	9.5	NA	NA	NA		NA	NA	CTGTTGAC+ACCTCAGT	5,617,260	1.00	97.81	2.19	1,696	100	93.55	35.88	NA	NA	NA
1	default	S5_Longley	6102487	NA	Healthy/Wild	MAU	M	A	Summer	Channel	2.9315068	9.7	NA	NA	NA		NA	NA	CGCAATCT+ACAGGCAT	2,937,660	1.00	97.71	2.29	887	100	93.75	35.91	NA	NA	NA
1	default	S5_Allen	6102488	NA	DFT1	MAU	M	A	Summer	Channel	1.9095890	6.7	NA	1	TRUE	deep	TRUE	55959.750	TTGCAGAC+ATCGTGGT	5,288,435	1.00	96.85	3.15	1,597	100	93.42	35.86	NA	NA	NA
1	default	S5_Cyan	6102489	NA	DFT1	MAW	M	A	Winter	Channel	3.3945205	9.1	NA	5	TRUE	deep	TRUE	206006.422	GTAGGAGT+GTCCTAAG	4,138,600	1.00	98.03	1.97	1,250	100	93.67	35.90	NA	NA	NA
1	default	S6_Myanmar	6102490	NA	Healthy	FAP	F	A	Spring	Channel	5.5726027	7.6	NA	NA	NA		NA	NA	GCACAACT+CTGAACGT	9,604,175	2.00	98.14	1.86	2,900	100	93.93	35.94	NA	NA	NA
1	default	S6_Pakana	6102491	NA	Healthy	MSP	M	S	Spring	Channel	1.6712329	7.8	NA	NA	NA		NA	NA	GTTACGCA+ATGGCGAT	4,383,521	1.00	98.09	1.91	1,324	100	93.90	35.95	NA	NA	NA
1	default	S6_Shelob	6102492	NA	DFT2	FAP	F	A	Spring	Channel	4.6219178	7.2	NA	NA	NA		NA	7455.435	GTCATCGA+GATCCACT	3,184,798	1.00	98.21	1.79	962	100	93.90	35.94	NA	NA	NA
1	default	S6_Stuttgart	6102493	NA	DFT1	MSP	M	S	Spring	Channel	0.5780822	12.7	NA	1	FALSE		TRUE	3809.806	AGCGTGTT+CTTCGCAA	3,056,475	1.00	97.50	2.50	923	100	93.95	35.95	NA	NA	NA
1	default	S6_Ben	6102494	NA	Wild	MAU	M	A	Summer	Channel	2.9150685	9.3	NA	NA	NA		NA	NA	AGAACGAG+ATTAGCCG	3,120,448	1.00	95.32	4.68	942	100	94.24	36.01	NA	NA	NA
1	default	S6_Hobbit	6102495	NA	Healthy	MSP	M	S	Spring	Channel	1.6273973	9	NA	NA	NA		NA	NA	ATATGCGC+ACGATCAG	7,342,414	2.00	97.00	3.00	2,217	100	93.93	35.96	NA	NA	NA
1	default	S7_Fredo	6102496	NA	Captive	MAA	M	A	Autumn	Healesville	2.9945205	8.8	NA	NA	NA		NA	NA	AGTGGATC+TCGATGAC	2,690,632	1.00	97.77	2.23	813	100	93.77	35.93	NA	NA	NA
1	default	S7_Piglet	6102497	NA	Captive	FSU	F	S	Summer	Healesville	1.5890411	4.76	NA	NA	NA		NA	NA	ATCACACG+GTAGCGTA	12,134,750	2.00	96.94	3.06	3,665	100	93.34	35.84	NA	NA	NA
1	default	S7_Violet	6102498	NA	DFT2	FSU	F	S	Summer	Channel	0.8739726	4.6	NA	1	TRUE	superficial	TRUE	5517.330	CTTGTCGA+CGATGTTC	2,768,682	1.00	98.19	1.81	836	100	93.85	35.93	NA	NA	NA
1	default	S7_Lieke	6102499	NA	Healthy/Wild	FSU	F	S	Summer	Channel	0.8657534	4.9	NA	NA	NA		NA	NA	TTCCAAGG+CTACAAGG	1,800,468	0.00	97.79	2.21	544	100	93.58	35.88	NA	NA	NA
1	default	S7_Willie	6102500	NA	DFT2	MAA	M	A	Autumn	Channel	2.1150685	8.2	NA	2	TRUE	deep	TRUE	1527.480	TGGACTCT+TCTTACGG	4,539,036	1.00	98.30	1.70	1,371	100	93.84	35.93	NA	NA	NA
1	default	S7_Smithers	6102501	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	3.1013699	8.9	NA	NA	NA		NA	NA	GCGTCATT+CAGACGTT	6,731,177	1.00	98.04	1.96	2,033	100	93.62	35.89	NA	NA	NA
1	default	S8_Orange	6102502	NA	DFT1	FAU	F	A	Summer	Channel	1.9150685	7.1	NA	7	TRUE	superficial	TRUE	43690.461	GTTGACCT+CAGTGCTT	5,072,309	1.00	97.92	2.08	1,532	100	93.36	35.84	NA	NA	NA
1	default	S8_Vinnie	6102503	NA	Healthy/Wild	MAU	M	A	Summer	Channel	2.9150685	8.8	NA	NA	NA		NA	NA	CGACGTTA+CTCTGGAT	7,901,958	2.00	98.02	1.98	2,386	100	94.09	35.98	NA	NA	NA
1	default	S8_Struggles	6102504	NA	Healthy/Wild	MAW	M	A	Winter	Channel	3.2821918	11.1	NA	NA	NA		NA	NA	ATCGATCG+TGGAAGCA	5,257,820	1.00	95.13	4.87	1,588	100	93.81	35.93	NA	NA	NA
1	default	S8_ST67	6102505	NA	DFT2	MAW	M	A	Winter	Channel	4.2794521	10.2	NA	1	TRUE	deep	TRUE	102246.432	TTGGACGT+GAAGTGCT	5,476,735	1.00	95.44	4.56	1,654	100	93.21	35.82	NA	NA	NA
1	default	S8_SusieQ	6102506	NA	Healthy/Wild	FAU	F	A	Summer	Channel	1.8712329	7.1	NA	NA	NA		NA	NA	CAGTCCAA+CGGTAATC	454,240	0.00	97.95	2.05	137	100	92.97	35.76	NA	NA	NA
1	default	S8_Yellow	6102507	NA	DFT2	MAU	M	A	Summer	Channel	1.8657534	10.5	NA	2	TRUE	deep	TRUE	139389.671	GAAGTTGG+CGAACAAC	6,469,324	1.00	97.99	2.01	1,954	100	94.00	35.97	NA	NA	NA
1	default	S9_Jane	6102508	NA	DFT1	FAA	F	A	Autumn	Channel	3.1972603	6.1	NA	NA	NA		NA	64646.036	GTTCTCGT+TGTTCCGT	12,489,105	3.00	97.92	2.08	3,772	100	93.54	35.88	NA	NA	NA
1	default	S9_Majula	6102509	NA	Healthy	FAA	F	A	Autumn	Channel	3.1150685	6	NA	NA	NA		NA	NA	ACTCTCGA+CTGTACCA	8,490,270	2.00	97.99	2.01	2,564	100	93.34	35.84	NA	NA	NA
1	default	S9_Snoopy	6102510	NA	Healthy/Wild	MSW	M	S	Winter	Channel	1.2794521	6.3	NA	NA	NA		NA	NA	AACTGAGC+CAATCAGG	1,709,078	0.00	98.04	1.96	516	100	93.06	35.78	NA	NA	NA
1	default	S9_Clytemnestra	6102511	NA	DFT1	FSP	F	S	Spring	Channel	1.6602740	6	NA	NA	NA		NA	34738.080	CATGGCTA+CACACATC	6,141,790	1.00	98.00	2.00	1,855	100	93.42	35.85	NA	NA	NA
1	default	S9_Zeus	6102512	NA	DFT2	MSW	M	S	Winter	Channel	1.3479452	7.3	NA	NA	NA		NA	100.000	CGTGTGTA+CCAACGAA	699,953	0.00	97.64	2.36	211	100	93.33	35.84	NA	NA	NA
1	default	S9_Sarabi	6102513	NA	Healthy/Wild	FSP	F	S	Spring	Channel	1.6328767	5.8	NA	NA	NA		NA	NA	CTTAGGAC+ACTCCTAC	3,945,235	1.00	98.11	1.89	1,191	100	93.55	35.88	NA	NA	NA
1	default	S10_McMahon	6102514	NA	Captive	MSP	M	S	Spring	Healesville	1.6493151	8.06	NA	NA	NA		NA	NA	AACGTCTG+AATGACGC	3,809,607	1.00	96.73	3.27	1,151	100	93.07	35.79	NA	NA	NA
1	default	S10_Logan	6102515	NA	Captive	MAU	M	A	Summer	Healesville	5.6000000	9.02	NA	NA	NA		NA	NA	GTGCCATA+CTCCTAGT	9,606,720	2.00	97.52	2.48	2,901	100	93.25	35.83	NA	NA	NA
1	default	S10_Laurasia	6102516	NA	DFT2	MSP	M	S	Spring	Channel	1.6383562	8	NA	NA	NA		NA	2697.152	TACGCTAC+CGTGTGAT	9,716,148	2.00	97.81	2.19	2,934	100	93.97	35.95	NA	NA	NA
1	default	S10_Enchilada	6102517	NA	DFT2	MAP	M	A	Spring	Channel	3.6438356	9.3	NA	14	FALSE		FALSE	301301.905	GATTGCTC+CACTTCAC	11,967,844	2.00	97.82	2.18	3,614	100	93.88	35.94	NA	NA	NA
1	default	S11_Baloo	6102518	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	3.1726027	11.5	NA	NA	NA		NA	NA	CTGATCGT+GCGCATAT	10,923,019	2.00	98.07	1.93	3,299	100	93.51	35.87	NA	NA	NA
1	default	S11_Murdoch/Washington	6102519	NA	DFT2	MAA	M	A	Autumn	Channel	1.4438356	8.9	NA	NA	NA		NA	127176.305	ACTGAGGT+GTCAACAG	2,003,509	0.00	97.82	2.18	605	100	93.39	35.85	NA	NA	NA
1	default	S11_Xuki	6102520	NA	Captive	MAA	M	A	Autumn	Monarto	2.0767123	7.95	NA	NA	NA		NA	NA	ATCGCCAT+TGCGTAAC	6,018,884	1.00	97.88	2.12	1,818	100	93.58	35.89	NA	NA	NA
1	default	S11_Kalamata	6102521	NA	Captive	MSW	M	S	Winter	Healesville	1.2849315	5.58	NA	NA	NA		NA	NA	GAACATCG+TCGACAAG	6,861,544	1.00	97.76	2.24	2,072	100	93.49	35.87	NA	NA	NA
1	default	S11_Freiburg	6102522	NA	N/A	MSW	M	S	Winter	Channel	1.2575342	8	NA	1	TRUE	deep	TRUE	0.000	GAATCCGA+AGCTAGTG	5,358,578	1.00	97.95	2.05	1,618	100	94.04	35.96	NA	NA	NA
1	default	S11_DanRhodes	6102523	NA	Healthy/Wild	MSW	M	S	Winter	Channel	1.3917808	6.8	NA	NA	NA		NA	NA	TCGCTGTT+CAAGTCGT	5,076,673	1.00	97.41	2.59	1,533	100	93.77	35.92	NA	NA	NA
1	default	S12_Abe	6102524	NA	DFT1	MAA	M	A	Autumn	Channel	3.1150685	7.9	NA	1	TRUE	deep	TRUE	13125.000	GTTGTTCG+CCAACTTC	6,446,467	1.00	97.60	2.40	1,947	100	93.65	35.90	NA	NA	NA
1	default	S12_Robin	6102525	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	2.1972603	7.7	NA	NA	NA		NA	NA	ATGACGTC+CTTCCTTC	6,756,342	1.00	97.12	2.88	2,040	100	92.53	35.68	NA	NA	NA
1	default	S12_Lilac	6102526	NA	DFT1	MAW	M	A	Winter	Channel	3.4109589	9.6	NA	5	TRUE	superficial	TRUE	37293.844	TATCGGTC+TACCGGAT	4,490,241	1.00	98.15	1.85	1,356	100	93.42	35.85	NA	NA	NA
1	default	S12_Eeyore	6102527	NA	Healthy/Wild	MAW	M	A	Winter	Channel	3.2794521	9	NA	NA	NA		NA	NA	TCAGGCTT+ATCATGCG	3,071,283	1.00	97.28	2.72	928	100	92.94	35.76	NA	NA	NA
1	default	S12_Wally	6102528	NA	Captive	MAA	M	A	Autumn	Monarto	2.1616438	N/A	NA	NA	NA		NA	NA	ACGACTTG+AACAGCGA	3,000,501	1.00	97.96	2.04	906	100	93.70	35.91	NA	NA	NA
1	default	S12_Naz	6102529	NA	Captive	MAW	M	A	Winter	Monarto	4.4547945	N/A	NA	NA	NA		NA	NA	CCTTGTAG+CCTTGGAA	5,927,942	1.00	97.40	2.60	1,790	100	93.86	35.94	NA	NA	NA
1	default	S13_Humbug	6102530	NA	Captive	MAU	M	A	Summer	Monarto	2.5397260	7.94	NA	NA	NA		NA	NA	GAGCAGTA+TGAGCTGT	1,702,608	0.00	98.07	1.93	514	100	93.58	35.89	NA	NA	NA
1	default	S13_Green	6102531	NA	DFT1	MAA	M	A	Autumn	Channel	3.1534247	7.5	NA	2	TRUE	superficial	TRUE	7407.222	GTCGAAGA+CCACATTG	2,623,695	1.00	97.38	2.62	792	100	92.94	35.76	NA	NA	NA
1	default	S13_Bump	6102532	NA	DFT1	MAU	M	A	Summer	Channel	3.9068493	9.3	NA	NA	NA		NA	629522.362	AGTTCGTC+ACTCGATC	9,435,053	2.00	98.16	1.84	2,849	100	93.64	35.89	NA	NA	NA
1	default	S13_Pangea	6102533	NA	Healthy/Wild	MAU	M	A	Summer	Channel	1.8794521	7.7	NA	NA	NA		NA	NA	ACCACGAT+GTCTGCAA	1,732,316	0.00	97.57	2.43	523	100	93.82	35.92	NA	NA	NA
1	default	S13_Finbar	6102534	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	2.1150685	6.8	NA	NA	NA		NA	NA	GAGACGAT+ACCGGTTA	1,152	0.00	96.70	3.30	0	100	92.06	35.56	NA	NA	NA
1	default	S13_Xerxes	6102535	NA	Captive	MAA	M	A	Autumn	Monarto	2.0767123	N/A	NA	NA	NA		NA	NA	GATAGGCT+CAGTCACA	7,628,257	2.00	98.34	1.66	2,304	100	93.62	35.89	NA	NA	NA
1	default	S14_Purple	6102536	NA	DFT1	FAW	F	A	Winter	Channel	2.3945205	6.8	NA	1	TRUE	deep	TRUE	24938.123	CGCTCTAT+TTGCAACG	5,432,182	1.00	97.81	2.19	1,641	100	93.93	35.95	NA	NA	NA
1	default	S14_Simba	6102537	NA	N/A	MAA	M	A	Autumn	Channel	3.1232877	9.4	NA	NA	NA		NA	NA	ATGCCTGT+AGATTGCG	11,175,146	2.00	97.91	2.09	3,375	100	93.71	35.90	NA	NA	NA
1	default	S14_Catani	6102538	NA	Captive	FAU	F	A	Summer	Healesville	2.8630137	6.46	NA	NA	NA		NA	NA	TGAAGACG+ACATGCCA	13,381,058	3.00	96.98	3.02	4,041	100	93.36	35.84	NA	NA	NA
2	default	S15_Halls	6102539	NA	Wild	FAU	F	A	Summer	Channel	1.9178082	4.7	NA	NA	NA		NA	NA	ATGACGTC+CTTCCTTC	1,308,213	0.28	97.48	2.52	395	100	93.03	35.79	NA	NA	NA
2	default	S15_Jessica	6102540	NA	Wild	FAP	F	A	Spring	Channel	3.6191781	7.4	NA	NA	NA		NA	NA	CATGGCTA+CACACATC	1,840,822	0.40	97.83	2.17	556	100	92.66	35.71	NA	NA	NA
2	default	S15_Amelia	6102541	NA	Captive	FAU	F	A	Summer	Healesville	4.7260274	5.85	NA	NA	NA		NA	NA	AACGTCTG+AATGACGC	8,041,818	1.73	96.32	3.68	2,429	100	92.18	35.63	NA	NA	NA
2	default	S15_London	6102542	NA	Captive	FAU	F	A	Summer	Healesville	3.7863014	4.8	NA	NA	NA		NA	NA	GTCATCGA+GATCCACT	5,758,111	1.24	96.72	3.28	1,739	100	91.87	35.55	NA	NA	NA
2	default	S15_Elyse	6102543	NA	Captive	FAP	F	A	Spring	Healesville	4.5232877	7.88	NA	NA	NA		NA	NA	AGCGTGTT+CTTCGCAA	5,774,637	1.24	96.76	3.24	1,744	100	91.56	35.50	NA	NA	NA
2	default	S15_Munchkin	6102544	NA	Wild	FAU	F	A	Summer	Channel	2.8739726	N/A	NA	NA	NA		NA	NA	AGAGTAGC+AAGGCGTA	7,216,024	1.56	97.99	2.01	2,179	100	93.47	35.87	NA	NA	NA
2	default	S16_Pelverata	6102545	NA	Wild	FAU	F	A	Summer	Channel	2.9287671	8.7	NA	NA	NA		NA	NA	CGCTCTAT+TTGCAACG	2,918,728	0.63	97.60	2.40	881	100	93.65	35.91	NA	NA	NA
2	default	S16_Lizzie	6102546	NA	Wild	FAU	F	A	Summer	Channel	1.9178082	6.8	NA	NA	NA		NA	NA	AGTTCGTC+ACTCGATC	10,758,280	2.32	97.92	2.08	3,249	100	93.20	35.82	NA	NA	NA
2	default	S16_Marley	6102547	NA	Captive	FAU	F	A	Summer	Monarto	4.7726027	7.51	NA	NA	NA		NA	NA	GCAAGATC+AGTCGAAG	348,621	0.08	97.24	2.76	105	100	92.77	35.73	NA	NA	NA
2	default	S16_Milla	6102548	NA	Captive	FAU	F	A	Summer	Healesville	2.7671233	5.82	NA	NA	NA		NA	NA	TTGCGAAG+AACACGCT	6,376,997	1.37	97.60	2.40	1,926	100	92.44	35.67	NA	NA	NA
2	default	S16_Maleficent	6102549	NA	Captive	FAU	F	A	Summer	Monarto	4.5589041	8.59	NA	NA	NA		NA	NA	TGCTTCCA+CGATCGAT	9,181,668	1.98	98.08	1.92	2,773	100	93.04	35.79	NA	NA	NA
2	default	S16_Lebanon	6102550	NA	Wild	FAU	F	A	Summer	Channel	5.8986301	9.8	NA	NA	NA		NA	NA	ATCCAGAG+TAACGTCG	7,154,214	1.54	73.27	26.73	2,161	100	92.95	35.78	NA	NA	NA
2	default	S17_Radiata	6102551	NA	Wild	FSU	F	S	Summer	Channel	0.8767123	3.7	NA	NA	NA		NA	NA	AACTGAGC+CAATCAGG	7,329,322	1.58	97.89	2.11	2,213	100	92.63	35.71	NA	NA	NA
2	default	S17_Shenzi	6102552	NA	Wild	FAU	F	A	Summer	Channel	1.8821918	7.6	NA	NA	NA		NA	NA	GTCGAAGA+CCACATTG	5,503,188	1.19	97.18	2.82	1,662	100	92.53	35.69	NA	NA	NA
2	default	S17_Anirak	6102553	NA	Captive	FSW	F	S	Winter	Healesville	1.3424658	5.42	NA	NA	NA		NA	NA	TGAGCTAG+GAACGGTT	3,740,878	0.81	97.15	2.85	1,130	100	93.03	35.79	NA	NA	NA
2	default	S17_Poly	6102554	NA	Captive	FSU	F	S	Summer	Monarto	1.7917808	6.67	NA	NA	NA		NA	NA	CTTAGGAC+ACTCCTAC	11,133,566	2.40	97.85	2.15	3,362	100	92.47	35.68	NA	NA	NA
2	default	S17_Vernicosa	6102555	NA	Wild	FSW	F	S	Winter	Channel	1.3643836	5.1	NA	NA	NA		NA	NA	CAATGTGG+TCTTCGAC	4,353,832	0.94	97.73	2.27	1,315	100	93.72	35.92	NA	NA	NA
2	default	S17_Methyl	6102556	NA	Captive	FSU	F	S	Summer	Healesville	0.7534247	2.72	NA	NA	NA		NA	NA	TGGTACAG+TCGAGAGT	7,887,375	1.70	97.57	2.43	2,382	100	93.15	35.81	NA	NA	NA
2	default	S18_Paprika	6102557	NA	Wild	FSW	F	S	Winter	Channel	1.2684932	6	NA	NA	NA		NA	NA	CGGTTGTT+GTGGTATG	11,961,885	2.58	95.85	4.15	3,612	100	92.97	35.78	NA	NA	NA
2	default	S18_Zara	6102558	NA	Captive	FSW	F	S	Winter	Monarto	1.5150685	4.6	NA	NA	NA		NA	NA	ACCACGAT+GTCTGCAA	7,308,907	1.58	97.70	2.30	2,207	100	92.84	35.75	NA	NA	NA
2	default	S18_Maze	6102559	NA	Captive	FSW	F	S	Winter	Healesville	1.4000000	5.82	NA	NA	NA		NA	NA	ACTGAGGT+GTCAACAG	2,984,930	0.64	97.59	2.41	901	100	92.49	35.69	NA	NA	NA
2	default	S18_Delores	6102560	NA	Captive	FSW	F	S	Winter	Healesville	1.2986301	5.1	NA	NA	NA		NA	NA	ACGACTTG+AACAGCGA	98	0.00	94.90	5.10	0	100	89.11	34.91	NA	NA	NA
2	default	S18_Haiko	6102561	NA	Wild	FSW	F	S	Winter	Channel	1.4082192	6.2	NA	NA	NA		NA	NA	GATCGAGT+GACGAACT	6,506,408	1.40	97.63	2.37	1,965	100	93.11	35.79	NA	NA	NA
2	default	S18_Hermione	6102562	NA	Wild	FSW	F	S	Winter	Channel	1.3424658	5.1	NA	NA	NA		NA	NA	CCTGATTG+GCTCAGTT	11,494,825	2.48	97.55	2.45	3,471	100	93.66	35.92	NA	NA	NA
2	default	S19_Tammy	6102563	NA	Wild	FSW	F	S	Winter	Channel	1.3671233	6	NA	NA	NA		NA	NA	ATCGATCG+TGGAAGCA	6,018,258	1.30	92.40	7.60	1,818	100	93.20	35.82	NA	NA	NA
2	default	S19_Morocco	6102564	NA	Captive	MAU	M	A	Summer	Monarto	4.5917808	9.82	NA	NA	NA		NA	NA	TCAGGCTT+ATCATGCG	4,879,629	1.05	97.38	2.62	1,474	100	92.63	35.71	NA	NA	NA
2	default	S19_Pecorino	6102565	NA	Captive	FSW	F	S	Winter	Healesville	1.1123288	6.02	NA	NA	NA		NA	NA	ATGCCTGT+AGATTGCG	4,857,395	1.05	97.58	2.42	1,467	100	92.64	35.72	NA	NA	NA
2	default	S19_Hooch	6102566	NA	Captive	MSA	M	S	Autumn	Healesville	1.0164384	6.1	NA	NA	NA		NA	NA	CCGTAAGA+AGAGTCCA	1,074,345	0.23	97.56	2.44	324	100	92.66	35.72	NA	NA	NA
2	default	S19_Darwin	6102567	NA	Wild	MAU	M	A	Summer	Channel	1.8849315	9	NA	NA	NA		NA	NA	CACCTGTT+AGGATAGC	0	0.00	NA	NA	0	NA	NA	NA	NA	NA	NA
2	default	S19_Tantor	6102568	NA	Wild	MSA	M	S	Autumn	Channel	1.1945205	6.9	NA	NA	NA		NA	NA	TGGACTCT+TCTTACGG	14,104,602	3.04	98.15	1.85	4,260	100	93.26	35.83	NA	NA	NA
2	default	S20_Muasy	6102569	NA	Captive	MSU	M	S	Summer	Healesville	1.4219178	6.62	NA	NA	NA		NA	NA	TTCCAAGG+CTACAAGG	10,669,237	2.30	97.64	2.36	3,222	100	92.79	35.74	NA	NA	NA
2	default	S20_Fuse	6102570	NA	Captive	MSU	M	S	Summer	Healesville	1.7315068	6.26	NA	NA	NA		NA	NA	GTAGGAGT+GTCCTAAG	10,603,179	2.29	97.34	2.66	3,202	100	92.01	35.58	NA	NA	NA
2	default	S20_Joseph	6102571	NA	Captive	MSU	M	S	Summer	Healesville	1.7479452	7.42	NA	NA	NA		NA	NA	CTTCGACT+GATCTTGC	5,642,025	1.22	96.37	3.63	1,704	100	91.87	35.56	NA	NA	NA
2	default	S20_Hudson	6102572	NA	Wild	MSU	M	S	Summer	Channel	0.9178082	4.1	NA	NA	NA		NA	NA	AAGCACTG+AGGTCAAC	4,225,812	0.91	97.88	2.12	1,276	100	93.16	35.81	NA	NA	NA
2	default	S20_Mace	6102573	NA	Wild	MSU	M	S	Summer	Channel	0.9068493	6	NA	NA	NA		NA	NA	CGTTGCAA+ATAGAGCG	6,385,393	1.38	95.82	4.18	1,928	100	92.04	35.61	NA	NA	NA
2	default	S20_Mint	6102574	NA	Wild	MSU	M	S	Summer	Channel	0.9068493	6.2	NA	NA	NA		NA	NA	GAAGGAAG+GACGTCAT	5,390,453	1.16	97.29	2.71	1,628	100	93.28	35.84	NA	NA	NA
2	default	S21_Sylvester	6102575	NA	Wild	MAW	M	A	Winter	Channel	2.2739726	8.5	NA	NA	NA		NA	NA	GTTGACCT+CAGTGCTT	5,450,970	1.18	97.62	2.38	1,646	100	93.04	35.79	NA	NA	NA
2	default	S21_Moama	6102576	NA	Captive	MAW	M	A	Winter	Monarto	1.6986301	8.75	NA	NA	NA		NA	NA	TTGGACGT+GAAGTGCT	10,821,895	2.33	96.02	3.98	3,268	100	92.72	35.73	NA	NA	NA
2	default	S21_Nauset	6102577	NA	Captive	MSW	M	S	Winter	Healesville	1.3041096	4.036	NA	NA	NA		NA	NA	TATCGGTC+TACCGGAT	8,693,920	1.87	98.13	1.87	2,626	100	92.32	35.65	NA	NA	NA
2	default	S21_Graham	6102578	NA	Captive	MSW	M	S	Winter	Healesville	1.3616438	6.4	NA	NA	NA		NA	NA	TACGCCTT+GCTACTCT	3,422,010	0.74	97.72	2.28	1,033	100	92.94	35.76	NA	NA	NA
2	default	S21_Harry	6102579	NA	Wild	MSW	M	S	Winter	Channel	1.3972603	7.2	NA	NA	NA		NA	NA	TTCGTTGG+TACACACG	10,720,172	2.31	97.18	2.82	3,237	100	92.73	35.72	NA	NA	NA
2	default	S21_Thor	6102580	NA	Wild	MSW	M	S	Winter	Channel	1.3452055	8.2	NA	NA	NA		NA	NA	GATGTGTG+TAGCCATG	6,051,140	1.30	97.63	2.37	1,827	100	93.21	35.83	NA	NA	NA
2	default	S22_Yanni	6102581	NA	Captive	MSW	M	S	Winter	Monarto	1.5178082	8.15	NA	NA	NA		NA	NA	CGTGTGTA+CCAACGAA	4,811,453	1.04	97.67	2.33	1,453	100	92.30	35.65	NA	NA	NA
2	default	S22_Gherkin	6102582	NA	Captive	MSW	M	S	Winter	Healesville	1.2630137	6.32	NA	NA	NA		NA	NA	CCTTGTAG+CCTTGGAA	5,913,586	1.27	97.05	2.95	1,786	100	92.92	35.77	NA	NA	NA
2	default	S22_Utah	6102583	NA	Wild	MSW	M	S	Winter	Channel	1.4438356	8.2	NA	NA	NA		NA	NA	ACTAGGAG+TATGGCAC	7,229,403	1.56	94.10	5.90	2,183	100	93.10	35.80	NA	NA	NA
2	default	S22_Stilton	6102584	NA	Captive	MSW	M	S	Winter	Healesville	1.2986301	5.76	NA	NA	NA		NA	NA	CGCAATCT+ACAGGCAT	8,306,537	1.79	96.94	3.06	2,509	100	91.86	35.55	NA	NA	NA
2	default	S22_Wellington	6102585	NA	Wild	MSW	M	S	Winter	Channel	1.4027397	6	NA	NA	NA		NA	NA	ACGGAACA+ACGAGAAC	7,783,546	1.68	97.13	2.87	2,351	100	92.96	35.78	NA	NA	NA
2	default	S22_Ginger	6102586	NA	Healthy/Wild	MSW	M	S	Winter	Channel	1.2712329	7.6	NA	NA	NA		NA	NA	ACAGCTCA+TACTGCTC	9,742,792	2.10	97.63	2.37	2,942	100	93.00	35.78	NA	NA	NA
2	default	S23_Meliza	6102587	NA	Ignore-Wild	FSW	F	S	Winter	Channel	1.4109589	7.3	NA	NA	NA		NA	NA	ACTCTCGA+CTGTACCA	6,770,476	1.46	97.75	2.25	2,045	100	93.04	35.79	NA	NA	NA
2	default	S23_Osiris	6102588	NA	Captive	MSW	M	S	Winter	Healesville	1.1753425	5.34	NA	NA	NA		NA	NA	AGTGGATC+TCGATGAC	3,528,471	0.76	97.24	2.76	1,066	100	93.08	35.79	NA	NA	NA
2	default	S23_X-boxx	6102589	NA	Captive	FSW	F	S	Winter	Monarto	1.4657534	4.34	NA	NA	NA		NA	NA	CTGTTGAC+ACCTCAGT	23	0.00	100.00	NaN	0	100	95.85	36.37	NA	NA	NA
2	default	S23_Quince	6102590	NA	Captive	MSW	M	S	Winter	Healesville	1.3835616	N/A	NA	NA	NA		NA	NA	AACCGTTC+CTAGCTCA	7,247,793	1.56	97.63	2.37	2,189	100	92.28	35.63	NA	NA	NA
2	default	S23_Saffron	6102591	NA	Wild	MSW	M	S	Winter	Channel	1.2164384	7.3	NA	NA	NA		NA	NA	GAAGTTGG+CGAACAAC	7,212,495	1.55	97.63	2.37	2,178	100	93.09	35.80	NA	NA	NA
2	default	S23_Gandalf	6102592	NA	Wild	MSW	M	S	Winter	Channel	1.3013699	6.4	NA	NA	NA		NA	NA	GATTGCTC+CACTTCAC	4,756,766	1.03	97.26	2.74	1,437	100	93.49	35.88	NA	NA	NA
2	default	S24_Boy	6102593	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	3.1452055	8.3	NA	NA	NA		NA	NA	CGACGTTA+CTCTGGAT	1,476,427	0.32	97.63	2.37	446	100	93.25	35.82	NA	NA	NA
2	default	S24_ST74	6102594	NA	Wild	MAW	M	A	Winter	Channel	4.2876712	10.4	NA	NA	NA		NA	NA	TGGTAGCT+AGGTGTTG	2,413,884	0.52	96.34	3.66	729	100	92.34	35.64	NA	NA	NA
2	default	S24_Malfoy	6102595	NA	Captive	MAW	M	A	Winter	Healesville	5.1041096	8.02	NA	NA	NA		NA	NA	TAACCGGT+ATCGTCTC	6,908,416	1.49	97.09	2.91	2,086	100	91.90	35.56	NA	NA	NA
2	default	S24_Licorice	6102596	NA	Captive	FSA	F	S	Autumn	Healesville	1.8246575	5.94	NA	NA	NA		NA	NA	TTGCAGAC+ATCGTGGT	4,475,461	0.96	95.05	4.95	1,352	100	91.42	35.46	NA	NA	NA
2	default	S24_Lisa	6102597	NA	Wild	FSA	F	S	Autumn	Channel	1.1013699	6.6	NA	NA	NA		NA	NA	CGCATGAT+AAGCCTGA	17,006,305	3.67	96.93	3.07	5,136	100	92.91	35.77	NA	NA	NA
2	default	S24_Shazza	6102598	NA	Healthy	FAW	F	A	Winter	Channel	2.0273973	6.2	NA	NA	NA		NA	NA	GAGCAGTA+TGAGCTGT	2,693,655	0.58	97.51	2.49	813	100	93.01	35.79	NA	NA	NA
2	default	S25_Tofu	6102599	NA	Captive	FAU	F	A	Summer	Healesville	2.7452055	N/A	NA	NA	NA		NA	NA	GAGACGAT+ACCGGTTA	12,024,421	2.59	95.37	4.63	3,631	100	93.06	35.79	NA	NA	NA
2	default	S25_Colette	6102600	NA	Captive	FAP	F	A	Spring	Healesville	6.6273973	7.52	NA	NA	NA		NA	NA	CTTGTCGA+CGATGTTC	10,751,797	2.32	97.49	2.51	3,247	100	92.77	35.74	NA	NA	NA
2	default	S25_CannonBall	6102601	NA	Captive	MSW	M	S	Winter	Healesville	1.1479452	6.54	NA	NA	NA		NA	NA	TACGCTAC+CGTGTGAT	13,015,081	2.81	97.36	2.64	3,931	100	92.61	35.70	NA	NA	NA
2	default	S25_Cornelius	6102602	NA	Captive	MAU	M	A	Summer	Healesville	3.6904110	7.74	NA	NA	NA		NA	NA	GAATCCGA+AGCTAGTG	3,336,983	0.72	97.10	2.90	1,008	100	93.09	35.79	NA	NA	NA
2	default	S25_Mardook	6102603	NA	Captive	MAW	M	A	Winter	Healesville	4.3013699	9.16	NA	NA	NA		NA	NA	GCACAACT+CTGAACGT	6,826,313	1.47	97.66	2.34	2,062	100	92.67	35.71	NA	NA	NA
2	default	S25_Negroni	6102604	NA	Captive	FSP	F	S	Spring	Healesville	1.6767123	5.86	NA	NA	NA		NA	NA	TGTGACTG+AGCCTATC	4,566,180	0.98	97.68	2.32	1,379	100	92.13	35.61	NA	NA	NA

MDS

# sample level
plotMDS(x)

# gene level
plotMDS(t(x))

DESeq2

ssHvDAV <- ss[which(grepl("Healthy",ss$Category) + grepl("DFT",ss$Category)>0),]
ssHvDAV

##     Lane Project              Client.ID  Sample  X     Category Sex.Age.Season
## 1      1 default          S1_LauraGrace 6102461 NA Healthy/Wild            FAP
## 3      1 default            S1_Sarafina 6102463 NA         DFT1            FAP
## 4      1 default                S1_Lucy 6102464 NA         DFT1            FAU
## 5      1 default            S1_Turmeric 6102465 NA Healthy/Wild            FAU
## 6      1 default           S1_Singapore 6102466 NA      Healthy            FAA
## 7      1 default             S2_Belinda 6102467 NA Healthy/Wild            FAU
## 10     1 default               S2_Rosie 6102470 NA Healthy/Wild            FAU
## 11     1 default                S2_Hope 6102471 NA         DFT1            FAU
## 12     1 default               S2_Sally 6102472 NA         DFT1            FAU
## 13     1 default                 S3_Apu 6102473 NA         DFT1            MAA
## 14     1 default              S3_Simone 6102474 NA         DFT1            FAW
## 15     1 default                S3_ST62 6102475 NA Healthy/Wild            FAW
## 18     1 default             S4_Kerchak 6102478 NA      Healthy            MAA
## 19     1 default             S4_Duffman 6102479 NA      Healthy            MAA
## 20     1 default              S4_Wiggum 6102480 NA         DFT1            MAA
## 21     1 default               S4_Hogan 6102481 NA      Healthy            MAA
## 22     1 default                S4_Pooh 6102482 NA         DFT1            MAA
## 23     1 default            S4_Milhouse 6102483 NA         DFT1            MAA
## 26     1 default               S5_Silky 6102486 NA Healthy/Wild            MAW
## 27     1 default             S5_Longley 6102487 NA Healthy/Wild            MAU
## 28     1 default               S5_Allen 6102488 NA         DFT1            MAU
## 29     1 default                S5_Cyan 6102489 NA         DFT1            MAW
## 30     1 default             S6_Myanmar 6102490 NA      Healthy            FAP
## 31     1 default              S6_Pakana 6102491 NA      Healthy            MSP
## 32     1 default              S6_Shelob 6102492 NA         DFT2            FAP
## 33     1 default           S6_Stuttgart 6102493 NA         DFT1            MSP
## 35     1 default              S6_Hobbit 6102495 NA      Healthy            MSP
## 38     1 default              S7_Violet 6102498 NA         DFT2            FSU
## 39     1 default               S7_Lieke 6102499 NA Healthy/Wild            FSU
## 40     1 default              S7_Willie 6102500 NA         DFT2            MAA
## 41     1 default            S7_Smithers 6102501 NA Healthy/Wild            MAA
## 42     1 default              S8_Orange 6102502 NA         DFT1            FAU
## 43     1 default              S8_Vinnie 6102503 NA Healthy/Wild            MAU
## 44     1 default           S8_Struggles 6102504 NA Healthy/Wild            MAW
## 45     1 default                S8_ST67 6102505 NA         DFT2            MAW
## 46     1 default              S8_SusieQ 6102506 NA Healthy/Wild            FAU
## 47     1 default              S8_Yellow 6102507 NA         DFT2            MAU
## 48     1 default                S9_Jane 6102508 NA         DFT1            FAA
## 49     1 default              S9_Majula 6102509 NA      Healthy            FAA
## 50     1 default              S9_Snoopy 6102510 NA Healthy/Wild            MSW
## 51     1 default        S9_Clytemnestra 6102511 NA         DFT1            FSP
## 52     1 default                S9_Zeus 6102512 NA         DFT2            MSW
## 53     1 default              S9_Sarabi 6102513 NA Healthy/Wild            FSP
## 56     1 default           S10_Laurasia 6102516 NA         DFT2            MSP
## 57     1 default          S10_Enchilada 6102517 NA         DFT2            MAP
## 58     1 default              S11_Baloo 6102518 NA Healthy/Wild            MAA
## 59     1 default S11_Murdoch/Washington 6102519 NA         DFT2            MAA
## 63     1 default          S11_DanRhodes 6102523 NA Healthy/Wild            MSW
## 64     1 default                S12_Abe 6102524 NA         DFT1            MAA
## 65     1 default              S12_Robin 6102525 NA Healthy/Wild            MAA
## 66     1 default              S12_Lilac 6102526 NA         DFT1            MAW
## 67     1 default             S12_Eeyore 6102527 NA Healthy/Wild            MAW
## 71     1 default              S13_Green 6102531 NA         DFT1            MAA
## 72     1 default               S13_Bump 6102532 NA         DFT1            MAU
## 73     1 default             S13_Pangea 6102533 NA Healthy/Wild            MAU
## 74     1 default             S13_Finbar 6102534 NA Healthy/Wild            MAA
## 76     1 default             S14_Purple 6102536 NA         DFT1            FAW
## 126    2 default             S22_Ginger 6102586 NA Healthy/Wild            MSW
## 133    2 default                S24_Boy 6102593 NA Healthy/Wild            MAA
## 138    2 default             S24_Shazza 6102598 NA      Healthy            FAW
##     Sex Age Season Location Agebyyear Weight X.1 Number.of.Tumours Ulcerated
## 1     F   A Spring  Channel 2.6219178    7.2  NA                NA        NA
## 3     F   A Spring  Channel 2.6191781    6.4  NA                NA        NA
## 4     F   A Summer  Channel 2.9178082    7.1  NA                 3      TRUE
## 5     F   A Summer  Channel 1.9095890    7.8  NA                NA        NA
## 6     F   A Autumn  Channel 5.0000000    8.6  NA                NA        NA
## 7     F   A Summer  Channel 1.8657534    6.6  NA                NA        NA
## 10    F   A Summer  Channel 1.8904110    7.4  NA                NA        NA
## 11    F   A Summer  Channel 2.9315068    6.4  NA                 4      TRUE
## 12    F   A Summer  Channel 2.9945205      7  NA                 1      TRUE
## 13    M   A Autumn  Channel 3.1150685    9.5  NA                 1     FALSE
## 14    F   A Winter  Channel 3.2739726   10.2  NA                 1      TRUE
## 15    F   A Winter  Channel 3.2712329   10.2  NA                NA        NA
## 18    M   A Autumn  Channel 4.1863014   11.6  NA                NA        NA
## 19    M   A Autumn  Channel 3.1123288   12.4  NA                NA        NA
## 20    M   A Autumn  Channel 3.1178082    8.3  NA                 2     FALSE
## 21    M   A Autumn  Channel 3.1616438   11.6  NA                NA        NA
## 22    M   A Autumn  Channel 4.1863014   11.8  NA                 1     FALSE
## 23    M   A Autumn  Channel 4.1013699     11  NA                 1      TRUE
## 26    M   A Winter  Channel 2.2849315    9.5  NA                NA        NA
## 27    M   A Summer  Channel 2.9315068    9.7  NA                NA        NA
## 28    M   A Summer  Channel 1.9095890    6.7  NA                 1      TRUE
## 29    M   A Winter  Channel 3.3945205    9.1  NA                 5      TRUE
## 30    F   A Spring  Channel 5.5726027    7.6  NA                NA        NA
## 31    M   S Spring  Channel 1.6712329    7.8  NA                NA        NA
## 32    F   A Spring  Channel 4.6219178    7.2  NA                NA        NA
## 33    M   S Spring  Channel 0.5780822   12.7  NA                 1     FALSE
## 35    M   S Spring  Channel 1.6273973      9  NA                NA        NA
## 38    F   S Summer  Channel 0.8739726    4.6  NA                 1      TRUE
## 39    F   S Summer  Channel 0.8657534    4.9  NA                NA        NA
## 40    M   A Autumn  Channel 2.1150685    8.2  NA                 2      TRUE
## 41    M   A Autumn  Channel 3.1013699    8.9  NA                NA        NA
## 42    F   A Summer  Channel 1.9150685    7.1  NA                 7      TRUE
## 43    M   A Summer  Channel 2.9150685    8.8  NA                NA        NA
## 44    M   A Winter  Channel 3.2821918   11.1  NA                NA        NA
## 45    M   A Winter  Channel 4.2794521   10.2  NA                 1      TRUE
## 46    F   A Summer  Channel 1.8712329    7.1  NA                NA        NA
## 47    M   A Summer  Channel 1.8657534   10.5  NA                 2      TRUE
## 48    F   A Autumn  Channel 3.1972603    6.1  NA                NA        NA
## 49    F   A Autumn  Channel 3.1150685      6  NA                NA        NA
## 50    M   S Winter  Channel 1.2794521    6.3  NA                NA        NA
## 51    F   S Spring  Channel 1.6602740      6  NA                NA        NA
## 52    M   S Winter  Channel 1.3479452    7.3  NA                NA        NA
## 53    F   S Spring  Channel 1.6328767    5.8  NA                NA        NA
## 56    M   S Spring  Channel 1.6383562      8  NA                NA        NA
## 57    M   A Spring  Channel 3.6438356    9.3  NA                14     FALSE
## 58    M   A Autumn  Channel 3.1726027   11.5  NA                NA        NA
## 59    M   A Autumn  Channel 1.4438356    8.9  NA                NA        NA
## 63    M   S Winter  Channel 1.3917808    6.8  NA                NA        NA
## 64    M   A Autumn  Channel 3.1150685    7.9  NA                 1      TRUE
## 65    M   A Autumn  Channel 2.1972603    7.7  NA                NA        NA
## 66    M   A Winter  Channel 3.4109589    9.6  NA                 5      TRUE
## 67    M   A Winter  Channel 3.2794521      9  NA                NA        NA
## 71    M   A Autumn  Channel 3.1534247    7.5  NA                 2      TRUE
## 72    M   A Summer  Channel 3.9068493    9.3  NA                NA        NA
## 73    M   A Summer  Channel 1.8794521    7.7  NA                NA        NA
## 74    M   A Autumn  Channel 2.1150685    6.8  NA                NA        NA
## 76    F   A Winter  Channel 2.3945205    6.8  NA                 1      TRUE
## 126   M   S Winter  Channel 1.2712329    7.6  NA                NA        NA
## 133   M   A Autumn  Channel 3.1452055    8.3  NA                NA        NA
## 138   F   A Winter  Channel 2.0273973    6.2  NA                NA        NA
##     UlcerationDepth Secondary.infection Total.tumour.volume  Barcode.sequence
## 1                                    NA                  NA AGAGTAGC+AAGGCGTA
## 3                                    NA           95140.790 ATCCAGAG+TAACGTCG
## 4              deep                TRUE          700825.170 AAGCACTG+AGGTCAAC
## 5                                    NA                  NA GAAGGAAG+GACGTCAT
## 6                                    NA                  NA ACAGCTCA+TACTGCTC
## 7                                    NA                  NA GCAAGATC+AGTCGAAG
## 10                                   NA                  NA CGCATGAT+AAGCCTGA
## 11             deep                TRUE          103815.530 AGCACTTC+ACGTCCAA
## 12             deep               FALSE            1588.272 CGTTGCAA+ATAGAGCG
## 13      superficial                TRUE             200.000 CCTTGATC+GATGGAGT
## 14      superficial                TRUE            2670.678 CGGTTGTT+GTGGTATG
## 15                                   NA                  NA TGAGCTAG+GAACGGTT
## 18                                   NA                  NA CGGCTAAT+CTCGTTCT
## 19                                   NA                  NA GACGATCT+TCGTGCAT
## 20                                FALSE            7272.309 GTGAAGTG+GAGCAATC
## 21                                   NA                  NA TACGCCTT+GCTACTCT
## 22                                FALSE            7577.856 CTTCGACT+GATCTTGC
## 23             deep                TRUE             480.000 AACCGTTC+CTAGCTCA
## 26                                   NA                  NA CTGTTGAC+ACCTCAGT
## 27                                   NA                  NA CGCAATCT+ACAGGCAT
## 28             deep                TRUE           55959.750 TTGCAGAC+ATCGTGGT
## 29             deep                TRUE          206006.422 GTAGGAGT+GTCCTAAG
## 30                                   NA                  NA GCACAACT+CTGAACGT
## 31                                   NA                  NA GTTACGCA+ATGGCGAT
## 32                                   NA            7455.435 GTCATCGA+GATCCACT
## 33                                 TRUE            3809.806 AGCGTGTT+CTTCGCAA
## 35                                   NA                  NA ATATGCGC+ACGATCAG
## 38      superficial                TRUE            5517.330 CTTGTCGA+CGATGTTC
## 39                                   NA                  NA TTCCAAGG+CTACAAGG
## 40             deep                TRUE            1527.480 TGGACTCT+TCTTACGG
## 41                                   NA                  NA GCGTCATT+CAGACGTT
## 42      superficial                TRUE           43690.461 GTTGACCT+CAGTGCTT
## 43                                   NA                  NA CGACGTTA+CTCTGGAT
## 44                                   NA                  NA ATCGATCG+TGGAAGCA
## 45             deep                TRUE          102246.432 TTGGACGT+GAAGTGCT
## 46                                   NA                  NA CAGTCCAA+CGGTAATC
## 47             deep                TRUE          139389.671 GAAGTTGG+CGAACAAC
## 48                                   NA           64646.036 GTTCTCGT+TGTTCCGT
## 49                                   NA                  NA ACTCTCGA+CTGTACCA
## 50                                   NA                  NA AACTGAGC+CAATCAGG
## 51                                   NA           34738.080 CATGGCTA+CACACATC
## 52                                   NA             100.000 CGTGTGTA+CCAACGAA
## 53                                   NA                  NA CTTAGGAC+ACTCCTAC
## 56                                   NA            2697.152 TACGCTAC+CGTGTGAT
## 57                                FALSE          301301.905 GATTGCTC+CACTTCAC
## 58                                   NA                  NA CTGATCGT+GCGCATAT
## 59                                   NA          127176.305 ACTGAGGT+GTCAACAG
## 63                                   NA                  NA TCGCTGTT+CAAGTCGT
## 64             deep                TRUE           13125.000 GTTGTTCG+CCAACTTC
## 65                                   NA                  NA ATGACGTC+CTTCCTTC
## 66      superficial                TRUE           37293.844 TATCGGTC+TACCGGAT
## 67                                   NA                  NA TCAGGCTT+ATCATGCG
## 71      superficial                TRUE            7407.222 GTCGAAGA+CCACATTG
## 72                                   NA          629522.362 AGTTCGTC+ACTCGATC
## 73                                   NA                  NA ACCACGAT+GTCTGCAA
## 74                                   NA                  NA GAGACGAT+ACCGGTTA
## 76             deep                TRUE           24938.123 CGCTCTAT+TTGCAACG
## 126                                  NA                  NA ACAGCTCA+TACTGCTC
## 133                                  NA                  NA CGACGTTA+CTCTGGAT
## 138                                  NA                  NA GAGCAGTA+TGAGCTGT
##     PF.Clusters X..of.the.lane X..Perfect.barcode X..One.mismatch.barcode
## 1     4,511,702           1.00              98.29                    1.71
## 3     6,491,488           1.00              74.07                   25.93
## 4     6,121,785           1.00              98.00                    2.00
## 5     6,479,047           1.00              97.28                    2.72
## 6     4,481,213           1.00              97.99                    2.01
## 7     3,296,547           1.00              97.40                    2.60
## 10    4,556,781           1.00              95.70                    4.30
## 11    1,971,776           0.00              97.96                    2.04
## 12    2,429,449           1.00              97.42                    2.58
## 13    4,939,580           1.00              96.94                    3.06
## 14    4,015,027           1.00              95.91                    4.09
## 15    4,579,681           1.00              96.67                    3.33
## 18    6,886,511           1.00              97.80                    2.20
## 19    8,301,336           2.00              97.75                    2.25
## 20    4,201,355           1.00              97.17                    2.83
## 21    2,990,345           1.00              97.77                    2.23
## 22    3,821,151           1.00              97.31                    2.69
## 23   10,492,900           2.00              98.13                    1.87
## 26    5,617,260           1.00              97.81                    2.19
## 27    2,937,660           1.00              97.71                    2.29
## 28    5,288,435           1.00              96.85                    3.15
## 29    4,138,600           1.00              98.03                    1.97
## 30    9,604,175           2.00              98.14                    1.86
## 31    4,383,521           1.00              98.09                    1.91
## 32    3,184,798           1.00              98.21                    1.79
## 33    3,056,475           1.00              97.50                    2.50
## 35    7,342,414           2.00              97.00                    3.00
## 38    2,768,682           1.00              98.19                    1.81
## 39    1,800,468           0.00              97.79                    2.21
## 40    4,539,036           1.00              98.30                    1.70
## 41    6,731,177           1.00              98.04                    1.96
## 42    5,072,309           1.00              97.92                    2.08
## 43    7,901,958           2.00              98.02                    1.98
## 44    5,257,820           1.00              95.13                    4.87
## 45    5,476,735           1.00              95.44                    4.56
## 46      454,240           0.00              97.95                    2.05
## 47    6,469,324           1.00              97.99                    2.01
## 48   12,489,105           3.00              97.92                    2.08
## 49    8,490,270           2.00              97.99                    2.01
## 50    1,709,078           0.00              98.04                    1.96
## 51    6,141,790           1.00              98.00                    2.00
## 52      699,953           0.00              97.64                    2.36
## 53    3,945,235           1.00              98.11                    1.89
## 56    9,716,148           2.00              97.81                    2.19
## 57   11,967,844           2.00              97.82                    2.18
## 58   10,923,019           2.00              98.07                    1.93
## 59    2,003,509           0.00              97.82                    2.18
## 63    5,076,673           1.00              97.41                    2.59
## 64    6,446,467           1.00              97.60                    2.40
## 65    6,756,342           1.00              97.12                    2.88
## 66    4,490,241           1.00              98.15                    1.85
## 67    3,071,283           1.00              97.28                    2.72
## 71    2,623,695           1.00              97.38                    2.62
## 72    9,435,053           2.00              98.16                    1.84
## 73    1,732,316           0.00              97.57                    2.43
## 74        1,152           0.00              96.70                    3.30
## 76    5,432,182           1.00              97.81                    2.19
## 126   9,742,792           2.10              97.63                    2.37
## 133   1,476,427           0.32              97.63                    2.37
## 138   2,693,655           0.58              97.51                    2.49
##     Yield..Mbases. X..PF.Cluster X.....Q30 Mean.Quality X.2 X.3 X.4
## 1            1,363           100     93.82        35.92  NA  NA  NA
## 3            1,960           100     93.74        35.92  NA  NA  NA
## 4            1,849           100     93.84        35.93  NA  NA  NA
## 5            1,957           100     93.85        35.94  NA  NA  NA
## 6            1,353           100     93.95        35.96  NA  NA  NA
## 7              996           100     94.12        35.98  NA  NA  NA
## 10           1,376           100     93.94        35.94  NA  NA  NA
## 11             595           100     93.62        35.88  NA  NA  NA
## 12             734           100     93.78        35.93  NA  NA  NA
## 13           1,492           100     93.71        35.91  NA  NA  NA
## 14           1,213           100     93.42        35.86  NA  NA  NA
## 15           1,383           100     93.75        35.92  NA  NA  NA
## 18           2,080           100     93.61        35.89  NA  NA  NA
## 19           2,507           100     93.83        35.93  NA  NA  NA
## 20           1,269           100     93.64        35.90  NA  NA  NA
## 21             903           100     93.71        35.90  NA  NA  NA
## 22           1,154           100     93.60        35.89  NA  NA  NA
## 23           3,169           100     93.80        35.92  NA  NA  NA
## 26           1,696           100     93.55        35.88  NA  NA  NA
## 27             887           100     93.75        35.91  NA  NA  NA
## 28           1,597           100     93.42        35.86  NA  NA  NA
## 29           1,250           100     93.67        35.90  NA  NA  NA
## 30           2,900           100     93.93        35.94  NA  NA  NA
## 31           1,324           100     93.90        35.95  NA  NA  NA
## 32             962           100     93.90        35.94  NA  NA  NA
## 33             923           100     93.95        35.95  NA  NA  NA
## 35           2,217           100     93.93        35.96  NA  NA  NA
## 38             836           100     93.85        35.93  NA  NA  NA
## 39             544           100     93.58        35.88  NA  NA  NA
## 40           1,371           100     93.84        35.93  NA  NA  NA
## 41           2,033           100     93.62        35.89  NA  NA  NA
## 42           1,532           100     93.36        35.84  NA  NA  NA
## 43           2,386           100     94.09        35.98  NA  NA  NA
## 44           1,588           100     93.81        35.93  NA  NA  NA
## 45           1,654           100     93.21        35.82  NA  NA  NA
## 46             137           100     92.97        35.76  NA  NA  NA
## 47           1,954           100     94.00        35.97  NA  NA  NA
## 48           3,772           100     93.54        35.88  NA  NA  NA
## 49           2,564           100     93.34        35.84  NA  NA  NA
## 50             516           100     93.06        35.78  NA  NA  NA
## 51           1,855           100     93.42        35.85  NA  NA  NA
## 52             211           100     93.33        35.84  NA  NA  NA
## 53           1,191           100     93.55        35.88  NA  NA  NA
## 56           2,934           100     93.97        35.95  NA  NA  NA
## 57           3,614           100     93.88        35.94  NA  NA  NA
## 58           3,299           100     93.51        35.87  NA  NA  NA
## 59             605           100     93.39        35.85  NA  NA  NA
## 63           1,533           100     93.77        35.92  NA  NA  NA
## 64           1,947           100     93.65        35.90  NA  NA  NA
## 65           2,040           100     92.53        35.68  NA  NA  NA
## 66           1,356           100     93.42        35.85  NA  NA  NA
## 67             928           100     92.94        35.76  NA  NA  NA
## 71             792           100     92.94        35.76  NA  NA  NA
## 72           2,849           100     93.64        35.89  NA  NA  NA
## 73             523           100     93.82        35.92  NA  NA  NA
## 74               0           100     92.06        35.56  NA  NA  NA
## 76           1,641           100     93.93        35.95  NA  NA  NA
## 126          2,942           100     93.00        35.78  NA  NA  NA
## 133            446           100     93.25        35.82  NA  NA  NA
## 138            813           100     93.01        35.79  NA  NA  NA

ssHvDAV <- ssHvDAV[ssHvDAV$Sample %in% colnames(x),]

dim(ssHvDAV)

## [1] 57 31

x1 <- x[,which(colnames(x) %in% ssHvDAV$Sample)]
dim(x1)

## [1] 310  57

x1 <- x1[which(rowMeans(x1)>10),]
dim(x1)

## [1] 259  57

ssHvDAV %>%  kbl(caption="Sample sheet")%>% kable_styling("hover",full_width=FALSE) 

Sample sheet

	Lane	Project	Client.ID	Sample	X	Category	Sex.Age.Season	Sex	Age	Season	Location	Agebyyear	Weight	X.1	Number.of.Tumours	Ulcerated	UlcerationDepth	Secondary.infection	Total.tumour.volume	Barcode.sequence	PF.Clusters	X..of.the.lane	X..Perfect.barcode	X..One.mismatch.barcode	Yield..Mbases.	X..PF.Cluster	X…..Q30	Mean.Quality	X.2	X.3	X.4
1	1	default	S1_LauraGrace	6102461	NA	Healthy/Wild	FAP	F	A	Spring	Channel	2.6219178	7.2	NA	NA	NA		NA	NA	AGAGTAGC+AAGGCGTA	4,511,702	1.00	98.29	1.71	1,363	100	93.82	35.92	NA	NA	NA
3	1	default	S1_Sarafina	6102463	NA	DFT1	FAP	F	A	Spring	Channel	2.6191781	6.4	NA	NA	NA		NA	95140.790	ATCCAGAG+TAACGTCG	6,491,488	1.00	74.07	25.93	1,960	100	93.74	35.92	NA	NA	NA
4	1	default	S1_Lucy	6102464	NA	DFT1	FAU	F	A	Summer	Channel	2.9178082	7.1	NA	3	TRUE	deep	TRUE	700825.170	AAGCACTG+AGGTCAAC	6,121,785	1.00	98.00	2.00	1,849	100	93.84	35.93	NA	NA	NA
5	1	default	S1_Turmeric	6102465	NA	Healthy/Wild	FAU	F	A	Summer	Channel	1.9095890	7.8	NA	NA	NA		NA	NA	GAAGGAAG+GACGTCAT	6,479,047	1.00	97.28	2.72	1,957	100	93.85	35.94	NA	NA	NA
6	1	default	S1_Singapore	6102466	NA	Healthy	FAA	F	A	Autumn	Channel	5.0000000	8.6	NA	NA	NA		NA	NA	ACAGCTCA+TACTGCTC	4,481,213	1.00	97.99	2.01	1,353	100	93.95	35.96	NA	NA	NA
7	1	default	S2_Belinda	6102467	NA	Healthy/Wild	FAU	F	A	Summer	Channel	1.8657534	6.6	NA	NA	NA		NA	NA	GCAAGATC+AGTCGAAG	3,296,547	1.00	97.40	2.60	996	100	94.12	35.98	NA	NA	NA
10	1	default	S2_Rosie	6102470	NA	Healthy/Wild	FAU	F	A	Summer	Channel	1.8904110	7.4	NA	NA	NA		NA	NA	CGCATGAT+AAGCCTGA	4,556,781	1.00	95.70	4.30	1,376	100	93.94	35.94	NA	NA	NA
11	1	default	S2_Hope	6102471	NA	DFT1	FAU	F	A	Summer	Channel	2.9315068	6.4	NA	4	TRUE	deep	TRUE	103815.530	AGCACTTC+ACGTCCAA	1,971,776	0.00	97.96	2.04	595	100	93.62	35.88	NA	NA	NA
12	1	default	S2_Sally	6102472	NA	DFT1	FAU	F	A	Summer	Channel	2.9945205	7	NA	1	TRUE	deep	FALSE	1588.272	CGTTGCAA+ATAGAGCG	2,429,449	1.00	97.42	2.58	734	100	93.78	35.93	NA	NA	NA
13	1	default	S3_Apu	6102473	NA	DFT1	MAA	M	A	Autumn	Channel	3.1150685	9.5	NA	1	FALSE	superficial	TRUE	200.000	CCTTGATC+GATGGAGT	4,939,580	1.00	96.94	3.06	1,492	100	93.71	35.91	NA	NA	NA
14	1	default	S3_Simone	6102474	NA	DFT1	FAW	F	A	Winter	Channel	3.2739726	10.2	NA	1	TRUE	superficial	TRUE	2670.678	CGGTTGTT+GTGGTATG	4,015,027	1.00	95.91	4.09	1,213	100	93.42	35.86	NA	NA	NA
15	1	default	S3_ST62	6102475	NA	Healthy/Wild	FAW	F	A	Winter	Channel	3.2712329	10.2	NA	NA	NA		NA	NA	TGAGCTAG+GAACGGTT	4,579,681	1.00	96.67	3.33	1,383	100	93.75	35.92	NA	NA	NA
18	1	default	S4_Kerchak	6102478	NA	Healthy	MAA	M	A	Autumn	Channel	4.1863014	11.6	NA	NA	NA		NA	NA	CGGCTAAT+CTCGTTCT	6,886,511	1.00	97.80	2.20	2,080	100	93.61	35.89	NA	NA	NA
19	1	default	S4_Duffman	6102479	NA	Healthy	MAA	M	A	Autumn	Channel	3.1123288	12.4	NA	NA	NA		NA	NA	GACGATCT+TCGTGCAT	8,301,336	2.00	97.75	2.25	2,507	100	93.83	35.93	NA	NA	NA
20	1	default	S4_Wiggum	6102480	NA	DFT1	MAA	M	A	Autumn	Channel	3.1178082	8.3	NA	2	FALSE		FALSE	7272.309	GTGAAGTG+GAGCAATC	4,201,355	1.00	97.17	2.83	1,269	100	93.64	35.90	NA	NA	NA
21	1	default	S4_Hogan	6102481	NA	Healthy	MAA	M	A	Autumn	Channel	3.1616438	11.6	NA	NA	NA		NA	NA	TACGCCTT+GCTACTCT	2,990,345	1.00	97.77	2.23	903	100	93.71	35.90	NA	NA	NA
22	1	default	S4_Pooh	6102482	NA	DFT1	MAA	M	A	Autumn	Channel	4.1863014	11.8	NA	1	FALSE		FALSE	7577.856	CTTCGACT+GATCTTGC	3,821,151	1.00	97.31	2.69	1,154	100	93.60	35.89	NA	NA	NA
23	1	default	S4_Milhouse	6102483	NA	DFT1	MAA	M	A	Autumn	Channel	4.1013699	11	NA	1	TRUE	deep	TRUE	480.000	AACCGTTC+CTAGCTCA	10,492,900	2.00	98.13	1.87	3,169	100	93.80	35.92	NA	NA	NA
26	1	default	S5_Silky	6102486	NA	Healthy/Wild	MAW	M	A	Winter	Channel	2.2849315	9.5	NA	NA	NA		NA	NA	CTGTTGAC+ACCTCAGT	5,617,260	1.00	97.81	2.19	1,696	100	93.55	35.88	NA	NA	NA
27	1	default	S5_Longley	6102487	NA	Healthy/Wild	MAU	M	A	Summer	Channel	2.9315068	9.7	NA	NA	NA		NA	NA	CGCAATCT+ACAGGCAT	2,937,660	1.00	97.71	2.29	887	100	93.75	35.91	NA	NA	NA
28	1	default	S5_Allen	6102488	NA	DFT1	MAU	M	A	Summer	Channel	1.9095890	6.7	NA	1	TRUE	deep	TRUE	55959.750	TTGCAGAC+ATCGTGGT	5,288,435	1.00	96.85	3.15	1,597	100	93.42	35.86	NA	NA	NA
29	1	default	S5_Cyan	6102489	NA	DFT1	MAW	M	A	Winter	Channel	3.3945205	9.1	NA	5	TRUE	deep	TRUE	206006.422	GTAGGAGT+GTCCTAAG	4,138,600	1.00	98.03	1.97	1,250	100	93.67	35.90	NA	NA	NA
30	1	default	S6_Myanmar	6102490	NA	Healthy	FAP	F	A	Spring	Channel	5.5726027	7.6	NA	NA	NA		NA	NA	GCACAACT+CTGAACGT	9,604,175	2.00	98.14	1.86	2,900	100	93.93	35.94	NA	NA	NA
31	1	default	S6_Pakana	6102491	NA	Healthy	MSP	M	S	Spring	Channel	1.6712329	7.8	NA	NA	NA		NA	NA	GTTACGCA+ATGGCGAT	4,383,521	1.00	98.09	1.91	1,324	100	93.90	35.95	NA	NA	NA
32	1	default	S6_Shelob	6102492	NA	DFT2	FAP	F	A	Spring	Channel	4.6219178	7.2	NA	NA	NA		NA	7455.435	GTCATCGA+GATCCACT	3,184,798	1.00	98.21	1.79	962	100	93.90	35.94	NA	NA	NA
33	1	default	S6_Stuttgart	6102493	NA	DFT1	MSP	M	S	Spring	Channel	0.5780822	12.7	NA	1	FALSE		TRUE	3809.806	AGCGTGTT+CTTCGCAA	3,056,475	1.00	97.50	2.50	923	100	93.95	35.95	NA	NA	NA
35	1	default	S6_Hobbit	6102495	NA	Healthy	MSP	M	S	Spring	Channel	1.6273973	9	NA	NA	NA		NA	NA	ATATGCGC+ACGATCAG	7,342,414	2.00	97.00	3.00	2,217	100	93.93	35.96	NA	NA	NA
38	1	default	S7_Violet	6102498	NA	DFT2	FSU	F	S	Summer	Channel	0.8739726	4.6	NA	1	TRUE	superficial	TRUE	5517.330	CTTGTCGA+CGATGTTC	2,768,682	1.00	98.19	1.81	836	100	93.85	35.93	NA	NA	NA
39	1	default	S7_Lieke	6102499	NA	Healthy/Wild	FSU	F	S	Summer	Channel	0.8657534	4.9	NA	NA	NA		NA	NA	TTCCAAGG+CTACAAGG	1,800,468	0.00	97.79	2.21	544	100	93.58	35.88	NA	NA	NA
40	1	default	S7_Willie	6102500	NA	DFT2	MAA	M	A	Autumn	Channel	2.1150685	8.2	NA	2	TRUE	deep	TRUE	1527.480	TGGACTCT+TCTTACGG	4,539,036	1.00	98.30	1.70	1,371	100	93.84	35.93	NA	NA	NA
41	1	default	S7_Smithers	6102501	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	3.1013699	8.9	NA	NA	NA		NA	NA	GCGTCATT+CAGACGTT	6,731,177	1.00	98.04	1.96	2,033	100	93.62	35.89	NA	NA	NA
42	1	default	S8_Orange	6102502	NA	DFT1	FAU	F	A	Summer	Channel	1.9150685	7.1	NA	7	TRUE	superficial	TRUE	43690.461	GTTGACCT+CAGTGCTT	5,072,309	1.00	97.92	2.08	1,532	100	93.36	35.84	NA	NA	NA
43	1	default	S8_Vinnie	6102503	NA	Healthy/Wild	MAU	M	A	Summer	Channel	2.9150685	8.8	NA	NA	NA		NA	NA	CGACGTTA+CTCTGGAT	7,901,958	2.00	98.02	1.98	2,386	100	94.09	35.98	NA	NA	NA
44	1	default	S8_Struggles	6102504	NA	Healthy/Wild	MAW	M	A	Winter	Channel	3.2821918	11.1	NA	NA	NA		NA	NA	ATCGATCG+TGGAAGCA	5,257,820	1.00	95.13	4.87	1,588	100	93.81	35.93	NA	NA	NA
45	1	default	S8_ST67	6102505	NA	DFT2	MAW	M	A	Winter	Channel	4.2794521	10.2	NA	1	TRUE	deep	TRUE	102246.432	TTGGACGT+GAAGTGCT	5,476,735	1.00	95.44	4.56	1,654	100	93.21	35.82	NA	NA	NA
47	1	default	S8_Yellow	6102507	NA	DFT2	MAU	M	A	Summer	Channel	1.8657534	10.5	NA	2	TRUE	deep	TRUE	139389.671	GAAGTTGG+CGAACAAC	6,469,324	1.00	97.99	2.01	1,954	100	94.00	35.97	NA	NA	NA
48	1	default	S9_Jane	6102508	NA	DFT1	FAA	F	A	Autumn	Channel	3.1972603	6.1	NA	NA	NA		NA	64646.036	GTTCTCGT+TGTTCCGT	12,489,105	3.00	97.92	2.08	3,772	100	93.54	35.88	NA	NA	NA
49	1	default	S9_Majula	6102509	NA	Healthy	FAA	F	A	Autumn	Channel	3.1150685	6	NA	NA	NA		NA	NA	ACTCTCGA+CTGTACCA	8,490,270	2.00	97.99	2.01	2,564	100	93.34	35.84	NA	NA	NA
50	1	default	S9_Snoopy	6102510	NA	Healthy/Wild	MSW	M	S	Winter	Channel	1.2794521	6.3	NA	NA	NA		NA	NA	AACTGAGC+CAATCAGG	1,709,078	0.00	98.04	1.96	516	100	93.06	35.78	NA	NA	NA
51	1	default	S9_Clytemnestra	6102511	NA	DFT1	FSP	F	S	Spring	Channel	1.6602740	6	NA	NA	NA		NA	34738.080	CATGGCTA+CACACATC	6,141,790	1.00	98.00	2.00	1,855	100	93.42	35.85	NA	NA	NA
53	1	default	S9_Sarabi	6102513	NA	Healthy/Wild	FSP	F	S	Spring	Channel	1.6328767	5.8	NA	NA	NA		NA	NA	CTTAGGAC+ACTCCTAC	3,945,235	1.00	98.11	1.89	1,191	100	93.55	35.88	NA	NA	NA
56	1	default	S10_Laurasia	6102516	NA	DFT2	MSP	M	S	Spring	Channel	1.6383562	8	NA	NA	NA		NA	2697.152	TACGCTAC+CGTGTGAT	9,716,148	2.00	97.81	2.19	2,934	100	93.97	35.95	NA	NA	NA
57	1	default	S10_Enchilada	6102517	NA	DFT2	MAP	M	A	Spring	Channel	3.6438356	9.3	NA	14	FALSE		FALSE	301301.905	GATTGCTC+CACTTCAC	11,967,844	2.00	97.82	2.18	3,614	100	93.88	35.94	NA	NA	NA
58	1	default	S11_Baloo	6102518	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	3.1726027	11.5	NA	NA	NA		NA	NA	CTGATCGT+GCGCATAT	10,923,019	2.00	98.07	1.93	3,299	100	93.51	35.87	NA	NA	NA
59	1	default	S11_Murdoch/Washington	6102519	NA	DFT2	MAA	M	A	Autumn	Channel	1.4438356	8.9	NA	NA	NA		NA	127176.305	ACTGAGGT+GTCAACAG	2,003,509	0.00	97.82	2.18	605	100	93.39	35.85	NA	NA	NA
63	1	default	S11_DanRhodes	6102523	NA	Healthy/Wild	MSW	M	S	Winter	Channel	1.3917808	6.8	NA	NA	NA		NA	NA	TCGCTGTT+CAAGTCGT	5,076,673	1.00	97.41	2.59	1,533	100	93.77	35.92	NA	NA	NA
64	1	default	S12_Abe	6102524	NA	DFT1	MAA	M	A	Autumn	Channel	3.1150685	7.9	NA	1	TRUE	deep	TRUE	13125.000	GTTGTTCG+CCAACTTC	6,446,467	1.00	97.60	2.40	1,947	100	93.65	35.90	NA	NA	NA
65	1	default	S12_Robin	6102525	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	2.1972603	7.7	NA	NA	NA		NA	NA	ATGACGTC+CTTCCTTC	6,756,342	1.00	97.12	2.88	2,040	100	92.53	35.68	NA	NA	NA
66	1	default	S12_Lilac	6102526	NA	DFT1	MAW	M	A	Winter	Channel	3.4109589	9.6	NA	5	TRUE	superficial	TRUE	37293.844	TATCGGTC+TACCGGAT	4,490,241	1.00	98.15	1.85	1,356	100	93.42	35.85	NA	NA	NA
67	1	default	S12_Eeyore	6102527	NA	Healthy/Wild	MAW	M	A	Winter	Channel	3.2794521	9	NA	NA	NA		NA	NA	TCAGGCTT+ATCATGCG	3,071,283	1.00	97.28	2.72	928	100	92.94	35.76	NA	NA	NA
71	1	default	S13_Green	6102531	NA	DFT1	MAA	M	A	Autumn	Channel	3.1534247	7.5	NA	2	TRUE	superficial	TRUE	7407.222	GTCGAAGA+CCACATTG	2,623,695	1.00	97.38	2.62	792	100	92.94	35.76	NA	NA	NA
72	1	default	S13_Bump	6102532	NA	DFT1	MAU	M	A	Summer	Channel	3.9068493	9.3	NA	NA	NA		NA	629522.362	AGTTCGTC+ACTCGATC	9,435,053	2.00	98.16	1.84	2,849	100	93.64	35.89	NA	NA	NA
73	1	default	S13_Pangea	6102533	NA	Healthy/Wild	MAU	M	A	Summer	Channel	1.8794521	7.7	NA	NA	NA		NA	NA	ACCACGAT+GTCTGCAA	1,732,316	0.00	97.57	2.43	523	100	93.82	35.92	NA	NA	NA
76	1	default	S14_Purple	6102536	NA	DFT1	FAW	F	A	Winter	Channel	2.3945205	6.8	NA	1	TRUE	deep	TRUE	24938.123	CGCTCTAT+TTGCAACG	5,432,182	1.00	97.81	2.19	1,641	100	93.93	35.95	NA	NA	NA
126	2	default	S22_Ginger	6102586	NA	Healthy/Wild	MSW	M	S	Winter	Channel	1.2712329	7.6	NA	NA	NA		NA	NA	ACAGCTCA+TACTGCTC	9,742,792	2.10	97.63	2.37	2,942	100	93.00	35.78	NA	NA	NA
133	2	default	S24_Boy	6102593	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	3.1452055	8.3	NA	NA	NA		NA	NA	CGACGTTA+CTCTGGAT	1,476,427	0.32	97.63	2.37	446	100	93.25	35.82	NA	NA	NA
138	2	default	S24_Shazza	6102598	NA	Healthy	FAW	F	A	Winter	Channel	2.0273973	6.2	NA	NA	NA		NA	NA	GAGCAGTA+TGAGCTGT	2,693,655	0.58	97.51	2.49	813	100	93.01	35.79	NA	NA	NA

col <- as.character(as.numeric(grepl("DFT",ssHvDAV$Category)))
col <- gsub("0","lightgreen",col)
col <- gsub("1","pink",col)

mds<-plotMDS(x1,pch=19,cex=3,col=col,main="MDS plot:Healthy and DFTD infected devils")
mtext("Healthy=green, DFTD=Red")

ssHvDAV$HelvDFT <- grepl("DFT",ssHvDAV$Category)
ssHvDAV$HelvDFT <- factor(ssHvDAV$HelvDFT)
ssHvDAV$MvF <- factor(ssHvDAV$Sex)
ssHvDAV$SubAdvAd <- as.numeric(factor(ssHvDAV$Age,levels=c("S","A"),ordered = TRUE))
ssHvDAV$Season <- factor(ssHvDAV$Season)
str(ssHvDAV)

## 'data.frame':    57 obs. of  34 variables:
##  $ Lane                   : int  1 1 1 1 1 1 1 1 1 1 ...
##  $ Project                : chr  "default" "default" "default" "default" ...
##  $ Client.ID              : chr  "S1_LauraGrace" "S1_Sarafina" "S1_Lucy" "S1_Turmeric" ...
##  $ Sample                 : int  6102461 6102463 6102464 6102465 6102466 6102467 6102470 6102471 6102472 6102473 ...
##  $ X                      : logi  NA NA NA NA NA NA ...
##  $ Category               : chr  "Healthy/Wild" "DFT1" "DFT1" "Healthy/Wild" ...
##  $ Sex.Age.Season         : chr  "FAP" "FAP" "FAU" "FAU" ...
##  $ Sex                    : chr  "F" "F" "F" "F" ...
##  $ Age                    : chr  "A" "A" "A" "A" ...
##  $ Season                 : Factor w/ 4 levels "Autumn","Spring",..: 2 2 3 3 1 3 3 3 3 1 ...
##  $ Location               : chr  "Channel" "Channel" "Channel" "Channel" ...
##  $ Agebyyear              : num  2.62 2.62 2.92 1.91 5 ...
##  $ Weight                 : chr  "7.2" "6.4" "7.1" "7.8" ...
##  $ X.1                    : logi  NA NA NA NA NA NA ...
##  $ Number.of.Tumours      : int  NA NA 3 NA NA NA NA 4 1 1 ...
##  $ Ulcerated              : logi  NA NA TRUE NA NA NA ...
##  $ UlcerationDepth        : chr  "" "" "deep" "" ...
##  $ Secondary.infection    : logi  NA NA TRUE NA NA NA ...
##  $ Total.tumour.volume    : num  NA 95141 700825 NA NA ...
##  $ Barcode.sequence       : chr  "AGAGTAGC+AAGGCGTA" "ATCCAGAG+TAACGTCG" "AAGCACTG+AGGTCAAC" "GAAGGAAG+GACGTCAT" ...
##  $ PF.Clusters            : chr  "4,511,702" "6,491,488" "6,121,785" "6,479,047" ...
##  $ X..of.the.lane         : num  1 1 1 1 1 1 1 0 1 1 ...
##  $ X..Perfect.barcode     : num  98.3 74.1 98 97.3 98 ...
##  $ X..One.mismatch.barcode: num  1.71 25.93 2 2.72 2.01 ...
##  $ Yield..Mbases.         : chr  "1,363" "1,960" "1,849" "1,957" ...
##  $ X..PF.Cluster          : int  100 100 100 100 100 100 100 100 100 100 ...
##  $ X.....Q30              : num  93.8 93.7 93.8 93.8 94 ...
##  $ Mean.Quality           : num  35.9 35.9 35.9 35.9 36 ...
##  $ X.2                    : logi  NA NA NA NA NA NA ...
##  $ X.3                    : logi  NA NA NA NA NA NA ...
##  $ X.4                    : logi  NA NA NA NA NA NA ...
##  $ HelvDFT                : Factor w/ 2 levels "FALSE","TRUE": 1 2 2 1 1 1 1 2 2 2 ...
##  $ MvF                    : Factor w/ 2 levels "F","M": 1 1 1 1 1 1 1 1 1 2 ...
##  $ SubAdvAd               : num  2 2 2 2 2 2 2 2 2 2 ...

dds <- DESeqDataSetFromMatrix(countData = x1 , colData = ssHvDAV, design = ~ Season + SubAdvAd + MvF + HelvDFT )

## converting counts to integer mode

##   the design formula contains one or more numeric variables with integer values,
##   specifying a model with increasing fold change for higher values.
##   did you mean for this to be a factor? if so, first convert
##   this variable to a factor using the factor() function

res <- DESeq(dds)

## estimating size factors

## estimating dispersions

## gene-wise dispersion estimates

## mean-dispersion relationship

## final dispersion estimates

## fitting model and testing

## -- replacing outliers and refitting for 3 genes
## -- DESeq argument 'minReplicatesForReplace' = 7 
## -- original counts are preserved in counts(dds)

## estimating dispersions

## fitting model and testing

z<- results(res)
vsd <- vst(dds, blind=FALSE,nsub=nrow(x1)-50)
zz<-cbind(as.data.frame(z),assay(vsd))
dge<-as.data.frame(zz[order(zz$pvalue),])
dgeHvDAV <-dge
sig <- subset(dge,padj<0.05)
sig1_up <- rownames(subset(sig,log2FoldChange>0))
sig1_dn <- rownames(subset(sig,log2FoldChange<0))
length(sig1_up)

## [1] 47

length(sig1_dn)

## [1] 38

maplot(dgeHvDAV,"Healthy vs DFTD infected devils")

make_volcano(dgeHvDAV,"Healthy vs DFTD infected devils")

col <- as.character(as.numeric(grepl("DFT",ssHvDAV$Category)))
col <- gsub("0","lightgreen",col)
col <- gsub("1","pink",col)

#The 100 in "sig 1:100, 1:6" can change to number of gene in the significant table
sig[1:100,1:6] %>%
  kbl(caption="Comparison of Healthy vs DFTD infected devils") %>%
  kable_paper("hover", full_width = T)

Comparison of Healthy vs DFTD infected devils

	baseMean	log2FoldChange	lfcSE	stat	pvalue	padj
IL18	1378.89171	1.4216987	0.2278241	6.240335	0.0000000	0.0000001
CD177	4036.44726	2.2675933	0.3955005	5.733478	0.0000000	0.0000013
SIRPA	230.15168	1.4403323	0.2700664	5.333253	0.0000001	0.0000083
TNFRSF1A	7937.40538	0.6563928	0.1244032	5.276334	0.0000001	0.0000085
IL18R1	1505.15210	0.7106447	0.1421932	4.997739	0.0000006	0.0000248
TLR6	2241.01267	0.6599388	0.1328564	4.967308	0.0000007	0.0000248
CXCR2	15626.02598	0.8829527	0.1780537	4.958912	0.0000007	0.0000248
CXCR1	13293.18040	0.8295716	0.1678130	4.943428	0.0000008	0.0000248
CATH3	41831.39927	1.9695546	0.4012904	4.908054	0.0000009	0.0000264
IL10Rb	357.91402	0.8703684	0.1816991	4.790165	0.0000017	0.0000430
CEBPA	1546.03019	0.8162664	0.1735768	4.702625	0.0000026	0.0000602
IL1B	7189.49256	0.7695187	0.1700345	4.525662	0.0000060	0.0001294
SELL	61542.23445	0.5666564	0.1297063	4.368765	0.0000125	0.0002480
TLR2	4878.38882	0.6994241	0.1642979	4.257048	0.0000207	0.0003817
CD14	47238.90486	0.7992779	0.1933159	4.134570	0.0000356	0.0006117
CCR2	5392.98323	0.7714379	0.1903951	4.051774	0.0000508	0.0008013
IgM	44362.50039	-0.7914067	0.1957534	-4.042875	0.0000528	0.0008013
CD40	1709.62513	-0.6188927	0.1552339	-3.986840	0.0000670	0.0009175
FCGR3	2004.31378	0.8128539	0.2039936	3.984704	0.0000676	0.0009175
Lgals3	15424.59804	0.5452460	0.1410398	3.865901	0.0001107	0.0014278
NOD2	1131.39181	0.9314895	0.2418769	3.851089	0.0001176	0.0014447
IgA	11875.23357	-0.8032157	0.2107693	-3.810876	0.0001385	0.0016239
NLRP3	7142.93996	0.6246242	0.1646698	3.793193	0.0001487	0.0016683
CSF1	1107.34083	0.9334962	0.2478123	3.766949	0.0001653	0.0017765
IL3Ra	5574.61380	0.6065126	0.1648544	3.679081	0.0002341	0.0024157
Tap2	13286.56312	-0.4281996	0.1183264	-3.618801	0.0002960	0.0029369
IL13Ra1	3129.07452	0.8650256	0.2404430	3.597632	0.0003211	0.0030686
MS4A1	5307.69234	-0.7278618	0.2030655	-3.584370	0.0003379	0.0031134
TLR4	1759.37709	0.7199813	0.2017320	3.569000	0.0003583	0.0031881
CCL20	21.72043	3.0066104	0.8679418	3.464069	0.0005321	0.0045758
IRF4	391.68295	-0.8209784	0.2447083	-3.354927	0.0007939	0.0066070
IL1RN	1792.35749	0.7901072	0.2383586	3.314784	0.0009171	0.0073944
CD244	1118.27272	-0.6869924	0.2088520	-3.289375	0.0010041	0.0077274
VEGF-A	181.78814	-0.7960213	0.2423643	-3.284401	0.0010220	0.0077274
IL21R	2132.21019	-0.6185323	0.1887360	-3.277235	0.0010483	0.0077274
KIT	1926.29479	-0.7498471	0.2360605	-3.176504	0.0014906	0.0104756
IDO1	926.99455	1.0603486	0.3346408	3.168617	0.0015317	0.0104756
CCL21	12.25281	-2.7704523	0.8749288	-3.166489	0.0015429	0.0104756
IL7RA	2601.36757	-0.5879152	0.1876978	-3.132244	0.0017348	0.0112587
cGAS	305.53336	0.5119667	0.1638031	3.125501	0.0017750	0.0112587
CCLD7	386.32632	-0.8248264	0.2640995	-3.123166	0.0017892	0.0112587
CD68	5148.75966	0.5270435	0.1706686	3.088111	0.0020143	0.0116402
IL5Ra	759.48031	-0.6883783	0.2229492	-3.087601	0.0020178	0.0116402
IL17RA	5030.35130	0.4178305	0.1353635	3.086728	0.0020237	0.0116402
GZMB	77.73320	-0.8442568	0.2735968	-3.085769	0.0020303	0.0116402
FCMR	23733.50231	-0.7326530	0.2406091	-3.044993	0.0023269	0.0130367
CD3e	9187.52833	-0.5089037	0.1674664	-3.038840	0.0023749	0.0130367
TLR9	549.28809	-0.4963321	0.1637184	-3.031621	0.0024324	0.0130744
CXCL16	4738.52901	0.8099241	0.2688476	3.012577	0.0025904	0.0136392
TLR8	4406.94924	0.4507645	0.1502565	2.999967	0.0027001	0.0139325
Spi1	40971.81009	0.5951804	0.2016170	2.952034	0.0031569	0.0159701
TNFSF13	504.93798	0.4468982	0.1517142	2.945658	0.0032227	0.0159895
CCR3	6015.04603	-0.6069859	0.2088550	-2.906255	0.0036578	0.0178060
CCLD5	307.67215	-0.7145179	0.2532231	-2.821693	0.0047771	0.0228239
TBX21	1030.11918	-0.6019088	0.2148120	-2.802026	0.0050783	0.0238218
IL4I0	109.48549	-1.1552694	0.4140615	-2.790091	0.0052693	0.0241643
CD3d	2941.03788	-0.5445226	0.1954596	-2.785858	0.0053386	0.0241643
STAT4	822.70369	-0.5962374	0.2153705	-2.768426	0.0056328	0.0250561
STAT5A	11504.87208	0.1993526	0.0724774	2.750547	0.0059496	0.0253656
CD22	7391.09535	-0.3975971	0.1446730	-2.748247	0.0059915	0.0253656
KLRK1	127.70924	-0.8030980	0.2922557	-2.747929	0.0059973	0.0253656
GUSB	3743.34687	0.3574039	0.1307987	2.732472	0.0062861	0.0261583
HPRT	1383.34298	0.4270392	0.1569867	2.720225	0.0065237	0.0265173
C3	50230.44991	0.3751699	0.1380576	2.717489	0.0065779	0.0265173
CXCR3	1268.03841	0.3301578	0.1225246	2.694626	0.0070468	0.0279703
IL11Ra	868.84154	-0.4293178	0.1611256	-2.664491	0.0077105	0.0301410
FCGR2B	758.37919	0.7344109	0.2762461	2.658538	0.0078480	0.0302208
PDCD1	838.36148	-0.5222280	0.1968652	-2.652719	0.0079846	0.0302947
CD37	23957.10438	0.2058992	0.0785338	2.621792	0.0087469	0.0327058
STAT1	4556.40471	-0.3647814	0.1398807	-2.607804	0.0091125	0.0335862
CD40LG	518.11160	-0.3817034	0.1474343	-2.588973	0.0096263	0.0347910
LTB	9370.96982	-0.5525602	0.2136720	-2.586021	0.0097091	0.0347910
MRC1	1697.86388	0.5573935	0.2169524	2.569198	0.0101934	0.0356522
NLCR5	5102.47684	-0.3591879	0.1398653	-2.568099	0.0102258	0.0356522
TCRa	4848.11526	-0.3827734	0.1510140	-2.534689	0.0112547	0.0386591
IgE	7126.86220	-0.6575153	0.2598296	-2.530563	0.0113880	0.0386591
CD19	2149.43290	-0.5530937	0.2201509	-2.512340	0.0119934	0.0401855
LAMP1	18568.59604	0.2741809	0.1101358	2.489480	0.0127930	0.0422512
FCER2	68.08283	0.8681775	0.3492985	2.485488	0.0129374	0.0422512
LTa	158.97038	-0.5176755	0.2089638	-2.477346	0.0132364	0.0426873
CD4	2011.71083	-0.2791259	0.1129334	-2.471597	0.0134511	0.0428443
FGF2	29.42641	-0.8305677	0.3382594	-2.455417	0.0140721	0.0442757
STAT3	5607.51291	0.4616450	0.1899225	2.430702	0.0150696	0.0466590
MME	57.45498	1.8071972	0.7443808	2.427786	0.0151913	0.0466590
TNFSF18	29.42332	2.0926135	0.8667106	2.414432	0.0157598	0.0478355
NA	NA	NA	NA	NA	NA	NA
NA.1	NA	NA	NA	NA	NA	NA
NA.2	NA	NA	NA	NA	NA	NA
NA.3	NA	NA	NA	NA	NA	NA
NA.4	NA	NA	NA	NA	NA	NA
NA.5	NA	NA	NA	NA	NA	NA
NA.6	NA	NA	NA	NA	NA	NA
NA.7	NA	NA	NA	NA	NA	NA
NA.8	NA	NA	NA	NA	NA	NA
NA.9	NA	NA	NA	NA	NA	NA
NA.10	NA	NA	NA	NA	NA	NA
NA.11	NA	NA	NA	NA	NA	NA
NA.12	NA	NA	NA	NA	NA	NA
NA.13	NA	NA	NA	NA	NA	NA
NA.14	NA	NA	NA	NA	NA	NA

write.table(dge,file="dgeHvDAV.tsv",sep="\t")

mx <- sig[,7:ncol(sig)]
mx <- head(mx,30)
colfunc <- colorRampPalette(c("blue", "white", "red"))
heatmap.2(as.matrix(mx),trace="none",scale="row",
          col=colfunc(25),ColSideColors=col,mar=c(5,12)) 

ssDFTAV <- ss[which(grepl("DFT1",ss$Category) + grepl("DFT2",ss$Category)>0),]
ssDFTAV

##    Lane Project              Client.ID  Sample  X Category Sex.Age.Season Sex
## 3     1 default            S1_Sarafina 6102463 NA     DFT1            FAP   F
## 4     1 default                S1_Lucy 6102464 NA     DFT1            FAU   F
## 11    1 default                S2_Hope 6102471 NA     DFT1            FAU   F
## 12    1 default               S2_Sally 6102472 NA     DFT1            FAU   F
## 13    1 default                 S3_Apu 6102473 NA     DFT1            MAA   M
## 14    1 default              S3_Simone 6102474 NA     DFT1            FAW   F
## 20    1 default              S4_Wiggum 6102480 NA     DFT1            MAA   M
## 22    1 default                S4_Pooh 6102482 NA     DFT1            MAA   M
## 23    1 default            S4_Milhouse 6102483 NA     DFT1            MAA   M
## 28    1 default               S5_Allen 6102488 NA     DFT1            MAU   M
## 29    1 default                S5_Cyan 6102489 NA     DFT1            MAW   M
## 32    1 default              S6_Shelob 6102492 NA     DFT2            FAP   F
## 33    1 default           S6_Stuttgart 6102493 NA     DFT1            MSP   M
## 38    1 default              S7_Violet 6102498 NA     DFT2            FSU   F
## 40    1 default              S7_Willie 6102500 NA     DFT2            MAA   M
## 42    1 default              S8_Orange 6102502 NA     DFT1            FAU   F
## 45    1 default                S8_ST67 6102505 NA     DFT2            MAW   M
## 47    1 default              S8_Yellow 6102507 NA     DFT2            MAU   M
## 48    1 default                S9_Jane 6102508 NA     DFT1            FAA   F
## 51    1 default        S9_Clytemnestra 6102511 NA     DFT1            FSP   F
## 52    1 default                S9_Zeus 6102512 NA     DFT2            MSW   M
## 56    1 default           S10_Laurasia 6102516 NA     DFT2            MSP   M
## 57    1 default          S10_Enchilada 6102517 NA     DFT2            MAP   M
## 59    1 default S11_Murdoch/Washington 6102519 NA     DFT2            MAA   M
## 64    1 default                S12_Abe 6102524 NA     DFT1            MAA   M
## 66    1 default              S12_Lilac 6102526 NA     DFT1            MAW   M
## 71    1 default              S13_Green 6102531 NA     DFT1            MAA   M
## 72    1 default               S13_Bump 6102532 NA     DFT1            MAU   M
## 76    1 default             S14_Purple 6102536 NA     DFT1            FAW   F
##    Age Season Location Agebyyear Weight X.1 Number.of.Tumours Ulcerated
## 3    A Spring  Channel 2.6191781    6.4  NA                NA        NA
## 4    A Summer  Channel 2.9178082    7.1  NA                 3      TRUE
## 11   A Summer  Channel 2.9315068    6.4  NA                 4      TRUE
## 12   A Summer  Channel 2.9945205      7  NA                 1      TRUE
## 13   A Autumn  Channel 3.1150685    9.5  NA                 1     FALSE
## 14   A Winter  Channel 3.2739726   10.2  NA                 1      TRUE
## 20   A Autumn  Channel 3.1178082    8.3  NA                 2     FALSE
## 22   A Autumn  Channel 4.1863014   11.8  NA                 1     FALSE
## 23   A Autumn  Channel 4.1013699     11  NA                 1      TRUE
## 28   A Summer  Channel 1.9095890    6.7  NA                 1      TRUE
## 29   A Winter  Channel 3.3945205    9.1  NA                 5      TRUE
## 32   A Spring  Channel 4.6219178    7.2  NA                NA        NA
## 33   S Spring  Channel 0.5780822   12.7  NA                 1     FALSE
## 38   S Summer  Channel 0.8739726    4.6  NA                 1      TRUE
## 40   A Autumn  Channel 2.1150685    8.2  NA                 2      TRUE
## 42   A Summer  Channel 1.9150685    7.1  NA                 7      TRUE
## 45   A Winter  Channel 4.2794521   10.2  NA                 1      TRUE
## 47   A Summer  Channel 1.8657534   10.5  NA                 2      TRUE
## 48   A Autumn  Channel 3.1972603    6.1  NA                NA        NA
## 51   S Spring  Channel 1.6602740      6  NA                NA        NA
## 52   S Winter  Channel 1.3479452    7.3  NA                NA        NA
## 56   S Spring  Channel 1.6383562      8  NA                NA        NA
## 57   A Spring  Channel 3.6438356    9.3  NA                14     FALSE
## 59   A Autumn  Channel 1.4438356    8.9  NA                NA        NA
## 64   A Autumn  Channel 3.1150685    7.9  NA                 1      TRUE
## 66   A Winter  Channel 3.4109589    9.6  NA                 5      TRUE
## 71   A Autumn  Channel 3.1534247    7.5  NA                 2      TRUE
## 72   A Summer  Channel 3.9068493    9.3  NA                NA        NA
## 76   A Winter  Channel 2.3945205    6.8  NA                 1      TRUE
##    UlcerationDepth Secondary.infection Total.tumour.volume  Barcode.sequence
## 3                                   NA           95140.790 ATCCAGAG+TAACGTCG
## 4             deep                TRUE          700825.170 AAGCACTG+AGGTCAAC
## 11            deep                TRUE          103815.530 AGCACTTC+ACGTCCAA
## 12            deep               FALSE            1588.272 CGTTGCAA+ATAGAGCG
## 13     superficial                TRUE             200.000 CCTTGATC+GATGGAGT
## 14     superficial                TRUE            2670.678 CGGTTGTT+GTGGTATG
## 20                               FALSE            7272.309 GTGAAGTG+GAGCAATC
## 22                               FALSE            7577.856 CTTCGACT+GATCTTGC
## 23            deep                TRUE             480.000 AACCGTTC+CTAGCTCA
## 28            deep                TRUE           55959.750 TTGCAGAC+ATCGTGGT
## 29            deep                TRUE          206006.422 GTAGGAGT+GTCCTAAG
## 32                                  NA            7455.435 GTCATCGA+GATCCACT
## 33                                TRUE            3809.806 AGCGTGTT+CTTCGCAA
## 38     superficial                TRUE            5517.330 CTTGTCGA+CGATGTTC
## 40            deep                TRUE            1527.480 TGGACTCT+TCTTACGG
## 42     superficial                TRUE           43690.461 GTTGACCT+CAGTGCTT
## 45            deep                TRUE          102246.432 TTGGACGT+GAAGTGCT
## 47            deep                TRUE          139389.671 GAAGTTGG+CGAACAAC
## 48                                  NA           64646.036 GTTCTCGT+TGTTCCGT
## 51                                  NA           34738.080 CATGGCTA+CACACATC
## 52                                  NA             100.000 CGTGTGTA+CCAACGAA
## 56                                  NA            2697.152 TACGCTAC+CGTGTGAT
## 57                               FALSE          301301.905 GATTGCTC+CACTTCAC
## 59                                  NA          127176.305 ACTGAGGT+GTCAACAG
## 64            deep                TRUE           13125.000 GTTGTTCG+CCAACTTC
## 66     superficial                TRUE           37293.844 TATCGGTC+TACCGGAT
## 71     superficial                TRUE            7407.222 GTCGAAGA+CCACATTG
## 72                                  NA          629522.362 AGTTCGTC+ACTCGATC
## 76            deep                TRUE           24938.123 CGCTCTAT+TTGCAACG
##    PF.Clusters X..of.the.lane X..Perfect.barcode X..One.mismatch.barcode
## 3    6,491,488              1              74.07                   25.93
## 4    6,121,785              1              98.00                    2.00
## 11   1,971,776              0              97.96                    2.04
## 12   2,429,449              1              97.42                    2.58
## 13   4,939,580              1              96.94                    3.06
## 14   4,015,027              1              95.91                    4.09
## 20   4,201,355              1              97.17                    2.83
## 22   3,821,151              1              97.31                    2.69
## 23  10,492,900              2              98.13                    1.87
## 28   5,288,435              1              96.85                    3.15
## 29   4,138,600              1              98.03                    1.97
## 32   3,184,798              1              98.21                    1.79
## 33   3,056,475              1              97.50                    2.50
## 38   2,768,682              1              98.19                    1.81
## 40   4,539,036              1              98.30                    1.70
## 42   5,072,309              1              97.92                    2.08
## 45   5,476,735              1              95.44                    4.56
## 47   6,469,324              1              97.99                    2.01
## 48  12,489,105              3              97.92                    2.08
## 51   6,141,790              1              98.00                    2.00
## 52     699,953              0              97.64                    2.36
## 56   9,716,148              2              97.81                    2.19
## 57  11,967,844              2              97.82                    2.18
## 59   2,003,509              0              97.82                    2.18
## 64   6,446,467              1              97.60                    2.40
## 66   4,490,241              1              98.15                    1.85
## 71   2,623,695              1              97.38                    2.62
## 72   9,435,053              2              98.16                    1.84
## 76   5,432,182              1              97.81                    2.19
##    Yield..Mbases. X..PF.Cluster X.....Q30 Mean.Quality X.2 X.3 X.4
## 3           1,960           100     93.74        35.92  NA  NA  NA
## 4           1,849           100     93.84        35.93  NA  NA  NA
## 11            595           100     93.62        35.88  NA  NA  NA
## 12            734           100     93.78        35.93  NA  NA  NA
## 13          1,492           100     93.71        35.91  NA  NA  NA
## 14          1,213           100     93.42        35.86  NA  NA  NA
## 20          1,269           100     93.64        35.90  NA  NA  NA
## 22          1,154           100     93.60        35.89  NA  NA  NA
## 23          3,169           100     93.80        35.92  NA  NA  NA
## 28          1,597           100     93.42        35.86  NA  NA  NA
## 29          1,250           100     93.67        35.90  NA  NA  NA
## 32            962           100     93.90        35.94  NA  NA  NA
## 33            923           100     93.95        35.95  NA  NA  NA
## 38            836           100     93.85        35.93  NA  NA  NA
## 40          1,371           100     93.84        35.93  NA  NA  NA
## 42          1,532           100     93.36        35.84  NA  NA  NA
## 45          1,654           100     93.21        35.82  NA  NA  NA
## 47          1,954           100     94.00        35.97  NA  NA  NA
## 48          3,772           100     93.54        35.88  NA  NA  NA
## 51          1,855           100     93.42        35.85  NA  NA  NA
## 52            211           100     93.33        35.84  NA  NA  NA
## 56          2,934           100     93.97        35.95  NA  NA  NA
## 57          3,614           100     93.88        35.94  NA  NA  NA
## 59            605           100     93.39        35.85  NA  NA  NA
## 64          1,947           100     93.65        35.90  NA  NA  NA
## 66          1,356           100     93.42        35.85  NA  NA  NA
## 71            792           100     92.94        35.76  NA  NA  NA
## 72          2,849           100     93.64        35.89  NA  NA  NA
## 76          1,641           100     93.93        35.95  NA  NA  NA

ssDFTAV <- ssDFTAV[ssDFTAV$Sample %in% colnames(x),]

dim(ssDFTAV)

## [1] 28 31

x1 <- x[,which(colnames(x) %in% ssDFTAV$Sample)]
dim(x1)

## [1] 310  28

x1 <- x1[which(rowMeans(x1)>10),]
dim(x1)

## [1] 258  28

ssDFTAV %>%  kbl(caption="Sample sheet")%>% kable_styling("hover",full_width=FALSE) 

Sample sheet

	Lane	Project	Client.ID	Sample	X	Category	Sex.Age.Season	Sex	Age	Season	Location	Agebyyear	Weight	X.1	Number.of.Tumours	Ulcerated	UlcerationDepth	Secondary.infection	Total.tumour.volume	Barcode.sequence	PF.Clusters	X..of.the.lane	X..Perfect.barcode	X..One.mismatch.barcode	Yield..Mbases.	X..PF.Cluster	X…..Q30	Mean.Quality	X.2	X.3	X.4
3	1	default	S1_Sarafina	6102463	NA	DFT1	FAP	F	A	Spring	Channel	2.6191781	6.4	NA	NA	NA		NA	95140.790	ATCCAGAG+TAACGTCG	6,491,488	1	74.07	25.93	1,960	100	93.74	35.92	NA	NA	NA
4	1	default	S1_Lucy	6102464	NA	DFT1	FAU	F	A	Summer	Channel	2.9178082	7.1	NA	3	TRUE	deep	TRUE	700825.170	AAGCACTG+AGGTCAAC	6,121,785	1	98.00	2.00	1,849	100	93.84	35.93	NA	NA	NA
11	1	default	S2_Hope	6102471	NA	DFT1	FAU	F	A	Summer	Channel	2.9315068	6.4	NA	4	TRUE	deep	TRUE	103815.530	AGCACTTC+ACGTCCAA	1,971,776	0	97.96	2.04	595	100	93.62	35.88	NA	NA	NA
12	1	default	S2_Sally	6102472	NA	DFT1	FAU	F	A	Summer	Channel	2.9945205	7	NA	1	TRUE	deep	FALSE	1588.272	CGTTGCAA+ATAGAGCG	2,429,449	1	97.42	2.58	734	100	93.78	35.93	NA	NA	NA
13	1	default	S3_Apu	6102473	NA	DFT1	MAA	M	A	Autumn	Channel	3.1150685	9.5	NA	1	FALSE	superficial	TRUE	200.000	CCTTGATC+GATGGAGT	4,939,580	1	96.94	3.06	1,492	100	93.71	35.91	NA	NA	NA
14	1	default	S3_Simone	6102474	NA	DFT1	FAW	F	A	Winter	Channel	3.2739726	10.2	NA	1	TRUE	superficial	TRUE	2670.678	CGGTTGTT+GTGGTATG	4,015,027	1	95.91	4.09	1,213	100	93.42	35.86	NA	NA	NA
20	1	default	S4_Wiggum	6102480	NA	DFT1	MAA	M	A	Autumn	Channel	3.1178082	8.3	NA	2	FALSE		FALSE	7272.309	GTGAAGTG+GAGCAATC	4,201,355	1	97.17	2.83	1,269	100	93.64	35.90	NA	NA	NA
22	1	default	S4_Pooh	6102482	NA	DFT1	MAA	M	A	Autumn	Channel	4.1863014	11.8	NA	1	FALSE		FALSE	7577.856	CTTCGACT+GATCTTGC	3,821,151	1	97.31	2.69	1,154	100	93.60	35.89	NA	NA	NA
23	1	default	S4_Milhouse	6102483	NA	DFT1	MAA	M	A	Autumn	Channel	4.1013699	11	NA	1	TRUE	deep	TRUE	480.000	AACCGTTC+CTAGCTCA	10,492,900	2	98.13	1.87	3,169	100	93.80	35.92	NA	NA	NA
28	1	default	S5_Allen	6102488	NA	DFT1	MAU	M	A	Summer	Channel	1.9095890	6.7	NA	1	TRUE	deep	TRUE	55959.750	TTGCAGAC+ATCGTGGT	5,288,435	1	96.85	3.15	1,597	100	93.42	35.86	NA	NA	NA
29	1	default	S5_Cyan	6102489	NA	DFT1	MAW	M	A	Winter	Channel	3.3945205	9.1	NA	5	TRUE	deep	TRUE	206006.422	GTAGGAGT+GTCCTAAG	4,138,600	1	98.03	1.97	1,250	100	93.67	35.90	NA	NA	NA
32	1	default	S6_Shelob	6102492	NA	DFT2	FAP	F	A	Spring	Channel	4.6219178	7.2	NA	NA	NA		NA	7455.435	GTCATCGA+GATCCACT	3,184,798	1	98.21	1.79	962	100	93.90	35.94	NA	NA	NA
33	1	default	S6_Stuttgart	6102493	NA	DFT1	MSP	M	S	Spring	Channel	0.5780822	12.7	NA	1	FALSE		TRUE	3809.806	AGCGTGTT+CTTCGCAA	3,056,475	1	97.50	2.50	923	100	93.95	35.95	NA	NA	NA
38	1	default	S7_Violet	6102498	NA	DFT2	FSU	F	S	Summer	Channel	0.8739726	4.6	NA	1	TRUE	superficial	TRUE	5517.330	CTTGTCGA+CGATGTTC	2,768,682	1	98.19	1.81	836	100	93.85	35.93	NA	NA	NA
40	1	default	S7_Willie	6102500	NA	DFT2	MAA	M	A	Autumn	Channel	2.1150685	8.2	NA	2	TRUE	deep	TRUE	1527.480	TGGACTCT+TCTTACGG	4,539,036	1	98.30	1.70	1,371	100	93.84	35.93	NA	NA	NA
42	1	default	S8_Orange	6102502	NA	DFT1	FAU	F	A	Summer	Channel	1.9150685	7.1	NA	7	TRUE	superficial	TRUE	43690.461	GTTGACCT+CAGTGCTT	5,072,309	1	97.92	2.08	1,532	100	93.36	35.84	NA	NA	NA
45	1	default	S8_ST67	6102505	NA	DFT2	MAW	M	A	Winter	Channel	4.2794521	10.2	NA	1	TRUE	deep	TRUE	102246.432	TTGGACGT+GAAGTGCT	5,476,735	1	95.44	4.56	1,654	100	93.21	35.82	NA	NA	NA
47	1	default	S8_Yellow	6102507	NA	DFT2	MAU	M	A	Summer	Channel	1.8657534	10.5	NA	2	TRUE	deep	TRUE	139389.671	GAAGTTGG+CGAACAAC	6,469,324	1	97.99	2.01	1,954	100	94.00	35.97	NA	NA	NA
48	1	default	S9_Jane	6102508	NA	DFT1	FAA	F	A	Autumn	Channel	3.1972603	6.1	NA	NA	NA		NA	64646.036	GTTCTCGT+TGTTCCGT	12,489,105	3	97.92	2.08	3,772	100	93.54	35.88	NA	NA	NA
51	1	default	S9_Clytemnestra	6102511	NA	DFT1	FSP	F	S	Spring	Channel	1.6602740	6	NA	NA	NA		NA	34738.080	CATGGCTA+CACACATC	6,141,790	1	98.00	2.00	1,855	100	93.42	35.85	NA	NA	NA
56	1	default	S10_Laurasia	6102516	NA	DFT2	MSP	M	S	Spring	Channel	1.6383562	8	NA	NA	NA		NA	2697.152	TACGCTAC+CGTGTGAT	9,716,148	2	97.81	2.19	2,934	100	93.97	35.95	NA	NA	NA
57	1	default	S10_Enchilada	6102517	NA	DFT2	MAP	M	A	Spring	Channel	3.6438356	9.3	NA	14	FALSE		FALSE	301301.905	GATTGCTC+CACTTCAC	11,967,844	2	97.82	2.18	3,614	100	93.88	35.94	NA	NA	NA
59	1	default	S11_Murdoch/Washington	6102519	NA	DFT2	MAA	M	A	Autumn	Channel	1.4438356	8.9	NA	NA	NA		NA	127176.305	ACTGAGGT+GTCAACAG	2,003,509	0	97.82	2.18	605	100	93.39	35.85	NA	NA	NA
64	1	default	S12_Abe	6102524	NA	DFT1	MAA	M	A	Autumn	Channel	3.1150685	7.9	NA	1	TRUE	deep	TRUE	13125.000	GTTGTTCG+CCAACTTC	6,446,467	1	97.60	2.40	1,947	100	93.65	35.90	NA	NA	NA
66	1	default	S12_Lilac	6102526	NA	DFT1	MAW	M	A	Winter	Channel	3.4109589	9.6	NA	5	TRUE	superficial	TRUE	37293.844	TATCGGTC+TACCGGAT	4,490,241	1	98.15	1.85	1,356	100	93.42	35.85	NA	NA	NA
71	1	default	S13_Green	6102531	NA	DFT1	MAA	M	A	Autumn	Channel	3.1534247	7.5	NA	2	TRUE	superficial	TRUE	7407.222	GTCGAAGA+CCACATTG	2,623,695	1	97.38	2.62	792	100	92.94	35.76	NA	NA	NA
72	1	default	S13_Bump	6102532	NA	DFT1	MAU	M	A	Summer	Channel	3.9068493	9.3	NA	NA	NA		NA	629522.362	AGTTCGTC+ACTCGATC	9,435,053	2	98.16	1.84	2,849	100	93.64	35.89	NA	NA	NA
76	1	default	S14_Purple	6102536	NA	DFT1	FAW	F	A	Winter	Channel	2.3945205	6.8	NA	1	TRUE	deep	TRUE	24938.123	CGCTCTAT+TTGCAACG	5,432,182	1	97.81	2.19	1,641	100	93.93	35.95	NA	NA	NA

col <- as.character(as.numeric(grepl("DFT2",ssDFTAV$Category)))
col <- gsub("0","red",col)
col <- gsub("1","darkred",col)

mds<-plotMDS(x1,pch=18,cex=4,col=col,main="MDS plot:DFT1 and DFT2 infected devils")
mtext("DFT1=Red, DFT2=dark Red")

ssDFTAV$DFTDStat <- grepl("DFT2",ssDFTAV$Category)
ssDFTAV$MvF <- factor(ssDFTAV$Sex)
ssDFTAV$SubAdvAd <- as.numeric(factor(ssDFTAV$Age,levels=c("S","A"),ordered = TRUE))
ssDFTAV$Season <- factor(ssDFTAV$Season)
str(ssDFTAV)

## 'data.frame':    28 obs. of  34 variables:
##  $ Lane                   : int  1 1 1 1 1 1 1 1 1 1 ...
##  $ Project                : chr  "default" "default" "default" "default" ...
##  $ Client.ID              : chr  "S1_Sarafina" "S1_Lucy" "S2_Hope" "S2_Sally" ...
##  $ Sample                 : int  6102463 6102464 6102471 6102472 6102473 6102474 6102480 6102482 6102483 6102488 ...
##  $ X                      : logi  NA NA NA NA NA NA ...
##  $ Category               : chr  "DFT1" "DFT1" "DFT1" "DFT1" ...
##  $ Sex.Age.Season         : chr  "FAP" "FAU" "FAU" "FAU" ...
##  $ Sex                    : chr  "F" "F" "F" "F" ...
##  $ Age                    : chr  "A" "A" "A" "A" ...
##  $ Season                 : Factor w/ 4 levels "Autumn","Spring",..: 2 3 3 3 1 4 1 1 1 3 ...
##  $ Location               : chr  "Channel" "Channel" "Channel" "Channel" ...
##  $ Agebyyear              : num  2.62 2.92 2.93 2.99 3.12 ...
##  $ Weight                 : chr  "6.4" "7.1" "6.4" "7" ...
##  $ X.1                    : logi  NA NA NA NA NA NA ...
##  $ Number.of.Tumours      : int  NA 3 4 1 1 1 2 1 1 1 ...
##  $ Ulcerated              : logi  NA TRUE TRUE TRUE FALSE TRUE ...
##  $ UlcerationDepth        : chr  "" "deep" "deep" "deep" ...
##  $ Secondary.infection    : logi  NA TRUE TRUE FALSE TRUE TRUE ...
##  $ Total.tumour.volume    : num  95141 700825 103816 1588 200 ...
##  $ Barcode.sequence       : chr  "ATCCAGAG+TAACGTCG" "AAGCACTG+AGGTCAAC" "AGCACTTC+ACGTCCAA" "CGTTGCAA+ATAGAGCG" ...
##  $ PF.Clusters            : chr  "6,491,488" "6,121,785" "1,971,776" "2,429,449" ...
##  $ X..of.the.lane         : num  1 1 0 1 1 1 1 1 2 1 ...
##  $ X..Perfect.barcode     : num  74.1 98 98 97.4 96.9 ...
##  $ X..One.mismatch.barcode: num  25.93 2 2.04 2.58 3.06 ...
##  $ Yield..Mbases.         : chr  "1,960" "1,849" "595" "734" ...
##  $ X..PF.Cluster          : int  100 100 100 100 100 100 100 100 100 100 ...
##  $ X.....Q30              : num  93.7 93.8 93.6 93.8 93.7 ...
##  $ Mean.Quality           : num  35.9 35.9 35.9 35.9 35.9 ...
##  $ X.2                    : logi  NA NA NA NA NA NA ...
##  $ X.3                    : logi  NA NA NA NA NA NA ...
##  $ X.4                    : logi  NA NA NA NA NA NA ...
##  $ DFTDStat               : logi  FALSE FALSE FALSE FALSE FALSE FALSE ...
##  $ MvF                    : Factor w/ 2 levels "F","M": 1 1 1 1 2 1 2 2 2 2 ...
##  $ SubAdvAd               : num  2 2 2 2 2 2 2 2 2 2 ...

dds <- DESeqDataSetFromMatrix(countData = x1 , colData = ssDFTAV, design = ~ Season + SubAdvAd + MvF + DFTDStat )

## converting counts to integer mode

##   the design formula contains one or more numeric variables with integer values,
##   specifying a model with increasing fold change for higher values.
##   did you mean for this to be a factor? if so, first convert
##   this variable to a factor using the factor() function

res <- DESeq(dds)

## estimating size factors

## estimating dispersions

## gene-wise dispersion estimates

## mean-dispersion relationship

## final dispersion estimates

## fitting model and testing

z<- results(res)
vsd <- vst(dds, blind=FALSE,nsub=nrow(x1)-50)
zz<-cbind(as.data.frame(z),assay(vsd))
dge<-as.data.frame(zz[order(zz$pvalue),])
dgeDFTAV <- dge
sig <- subset(dge,padj<0.05)
sig1_up <- rownames(subset(sig,log2FoldChange>0))
sig1_dn <- rownames(subset(sig,log2FoldChange<0))
length(sig1_up)

## [1] 0

length(sig1_dn)

## [1] 0

maplot(dgeDFTAV,"DFT1 vs DFT2 infected devils")

make_volcano(dgeDFTAV,"DFT1 vs DFT2 infected devils")

ssWvCAV <- ss[which(grepl("Wild",ss$Category) + grepl("Captive",ss$Category)>0),]
ssWvCAV

##     Lane Project      Client.ID  Sample  X     Category Sex.Age.Season Sex Age
## 1      1 default  S1_LauraGrace 6102461 NA Healthy/Wild            FAP   F   A
## 5      1 default    S1_Turmeric 6102465 NA Healthy/Wild            FAU   F   A
## 7      1 default     S2_Belinda 6102467 NA Healthy/Wild            FAU   F   A
## 8      1 default      S2_B99008 6102468 NA      Captive            FAU   F   A
## 9      1 default      S2_Bonney 6102469 NA      Captive            FAU   F   A
## 10     1 default       S2_Rosie 6102470 NA Healthy/Wild            FAU   F   A
## 15     1 default        S3_ST62 6102475 NA Healthy/Wild            FAW   F   A
## 16     1 default      S3_Jessie 6102476 NA      Captive            FAW   F   A
## 17     1 default    S3_McCauley 6102477 NA      Captive            MAA   M   A
## 24     1 default      S5_Fenton 6102484 NA      Captive            MAU   M   A
## 25     1 default S5_Bryan Adams 6102485 NA      Captive            MAW   M   A
## 26     1 default       S5_Silky 6102486 NA Healthy/Wild            MAW   M   A
## 27     1 default     S5_Longley 6102487 NA Healthy/Wild            MAU   M   A
## 34     1 default         S6_Ben 6102494 NA         Wild            MAU   M   A
## 36     1 default       S7_Fredo 6102496 NA      Captive            MAA   M   A
## 37     1 default      S7_Piglet 6102497 NA      Captive            FSU   F   S
## 39     1 default       S7_Lieke 6102499 NA Healthy/Wild            FSU   F   S
## 41     1 default    S7_Smithers 6102501 NA Healthy/Wild            MAA   M   A
## 43     1 default      S8_Vinnie 6102503 NA Healthy/Wild            MAU   M   A
## 44     1 default   S8_Struggles 6102504 NA Healthy/Wild            MAW   M   A
## 46     1 default      S8_SusieQ 6102506 NA Healthy/Wild            FAU   F   A
## 50     1 default      S9_Snoopy 6102510 NA Healthy/Wild            MSW   M   S
## 53     1 default      S9_Sarabi 6102513 NA Healthy/Wild            FSP   F   S
## 54     1 default    S10_McMahon 6102514 NA      Captive            MSP   M   S
## 55     1 default      S10_Logan 6102515 NA      Captive            MAU   M   A
## 58     1 default      S11_Baloo 6102518 NA Healthy/Wild            MAA   M   A
## 60     1 default       S11_Xuki 6102520 NA      Captive            MAA   M   A
## 61     1 default   S11_Kalamata 6102521 NA      Captive            MSW   M   S
## 63     1 default  S11_DanRhodes 6102523 NA Healthy/Wild            MSW   M   S
## 65     1 default      S12_Robin 6102525 NA Healthy/Wild            MAA   M   A
## 67     1 default     S12_Eeyore 6102527 NA Healthy/Wild            MAW   M   A
## 68     1 default      S12_Wally 6102528 NA      Captive            MAA   M   A
## 69     1 default        S12_Naz 6102529 NA      Captive            MAW   M   A
## 70     1 default     S13_Humbug 6102530 NA      Captive            MAU   M   A
## 73     1 default     S13_Pangea 6102533 NA Healthy/Wild            MAU   M   A
## 74     1 default     S13_Finbar 6102534 NA Healthy/Wild            MAA   M   A
## 75     1 default     S13_Xerxes 6102535 NA      Captive            MAA   M   A
## 78     1 default     S14_Catani 6102538 NA      Captive            FAU   F   A
## 79     2 default      S15_Halls 6102539 NA         Wild            FAU   F   A
## 80     2 default    S15_Jessica 6102540 NA         Wild            FAP   F   A
## 81     2 default     S15_Amelia 6102541 NA      Captive            FAU   F   A
## 82     2 default     S15_London 6102542 NA      Captive            FAU   F   A
## 83     2 default      S15_Elyse 6102543 NA      Captive            FAP   F   A
## 84     2 default   S15_Munchkin 6102544 NA         Wild            FAU   F   A
## 85     2 default  S16_Pelverata 6102545 NA         Wild            FAU   F   A
## 86     2 default     S16_Lizzie 6102546 NA         Wild            FAU   F   A
## 87     2 default     S16_Marley 6102547 NA      Captive            FAU   F   A
## 88     2 default      S16_Milla 6102548 NA      Captive            FAU   F   A
## 89     2 default S16_Maleficent 6102549 NA      Captive            FAU   F   A
## 90     2 default    S16_Lebanon 6102550 NA         Wild            FAU   F   A
## 91     2 default    S17_Radiata 6102551 NA         Wild            FSU   F   S
## 92     2 default     S17_Shenzi 6102552 NA         Wild            FAU   F   A
## 93     2 default     S17_Anirak 6102553 NA      Captive            FSW   F   S
## 94     2 default       S17_Poly 6102554 NA      Captive            FSU   F   S
## 95     2 default  S17_Vernicosa 6102555 NA         Wild            FSW   F   S
## 96     2 default     S17_Methyl 6102556 NA      Captive            FSU   F   S
## 97     2 default    S18_Paprika 6102557 NA         Wild            FSW   F   S
## 98     2 default       S18_Zara 6102558 NA      Captive            FSW   F   S
## 99     2 default       S18_Maze 6102559 NA      Captive            FSW   F   S
## 100    2 default    S18_Delores 6102560 NA      Captive            FSW   F   S
## 101    2 default      S18_Haiko 6102561 NA         Wild            FSW   F   S
## 102    2 default   S18_Hermione 6102562 NA         Wild            FSW   F   S
## 103    2 default      S19_Tammy 6102563 NA         Wild            FSW   F   S
## 104    2 default    S19_Morocco 6102564 NA      Captive            MAU   M   A
## 105    2 default   S19_Pecorino 6102565 NA      Captive            FSW   F   S
## 106    2 default      S19_Hooch 6102566 NA      Captive            MSA   M   S
## 107    2 default     S19_Darwin 6102567 NA         Wild            MAU   M   A
## 108    2 default     S19_Tantor 6102568 NA         Wild            MSA   M   S
## 109    2 default      S20_Muasy 6102569 NA      Captive            MSU   M   S
## 110    2 default       S20_Fuse 6102570 NA      Captive            MSU   M   S
## 111    2 default     S20_Joseph 6102571 NA      Captive            MSU   M   S
## 112    2 default     S20_Hudson 6102572 NA         Wild            MSU   M   S
## 113    2 default       S20_Mace 6102573 NA         Wild            MSU   M   S
## 114    2 default       S20_Mint 6102574 NA         Wild            MSU   M   S
## 115    2 default  S21_Sylvester 6102575 NA         Wild            MAW   M   A
## 116    2 default      S21_Moama 6102576 NA      Captive            MAW   M   A
## 117    2 default     S21_Nauset 6102577 NA      Captive            MSW   M   S
## 118    2 default     S21_Graham 6102578 NA      Captive            MSW   M   S
## 119    2 default      S21_Harry 6102579 NA         Wild            MSW   M   S
## 120    2 default       S21_Thor 6102580 NA         Wild            MSW   M   S
## 121    2 default      S22_Yanni 6102581 NA      Captive            MSW   M   S
## 122    2 default    S22_Gherkin 6102582 NA      Captive            MSW   M   S
## 123    2 default       S22_Utah 6102583 NA         Wild            MSW   M   S
## 124    2 default    S22_Stilton 6102584 NA      Captive            MSW   M   S
## 125    2 default S22_Wellington 6102585 NA         Wild            MSW   M   S
## 126    2 default     S22_Ginger 6102586 NA Healthy/Wild            MSW   M   S
## 127    2 default     S23_Meliza 6102587 NA  Ignore-Wild            FSW   F   S
## 128    2 default     S23_Osiris 6102588 NA      Captive            MSW   M   S
## 129    2 default     S23_X-boxx 6102589 NA      Captive            FSW   F   S
## 130    2 default     S23_Quince 6102590 NA      Captive            MSW   M   S
## 131    2 default    S23_Saffron 6102591 NA         Wild            MSW   M   S
## 132    2 default    S23_Gandalf 6102592 NA         Wild            MSW   M   S
## 133    2 default        S24_Boy 6102593 NA Healthy/Wild            MAA   M   A
## 134    2 default       S24_ST74 6102594 NA         Wild            MAW   M   A
## 135    2 default     S24_Malfoy 6102595 NA      Captive            MAW   M   A
## 136    2 default   S24_Licorice 6102596 NA      Captive            FSA   F   S
## 137    2 default       S24_Lisa 6102597 NA         Wild            FSA   F   S
## 139    2 default       S25_Tofu 6102599 NA      Captive            FAU   F   A
## 140    2 default    S25_Colette 6102600 NA      Captive            FAP   F   A
## 141    2 default S25_CannonBall 6102601 NA      Captive            MSW   M   S
## 142    2 default  S25_Cornelius 6102602 NA      Captive            MAU   M   A
## 143    2 default    S25_Mardook 6102603 NA      Captive            MAW   M   A
## 144    2 default    S25_Negroni 6102604 NA      Captive            FSP   F   S
##     Season    Location Agebyyear Weight X.1 Number.of.Tumours Ulcerated
## 1   Spring     Channel 2.6219178    7.2  NA                NA        NA
## 5   Summer     Channel 1.9095890    7.8  NA                NA        NA
## 7   Summer     Channel 1.8657534    6.6  NA                NA        NA
## 8   Summer     Monarto 3.7369863   8.12  NA                NA        NA
## 9   Summer Healesville 2.5863014    5.4  NA                NA        NA
## 10  Summer     Channel 1.8904110    7.4  NA                NA        NA
## 15  Winter     Channel 3.2712329   10.2  NA                NA        NA
## 16  Winter     Monarto 2.1945205  12.95  NA                NA        NA
## 17  Autumn Healesville 3.0328767   8.52  NA                NA        NA
## 24  Summer     Monarto 3.0000000    N/A  NA                NA        NA
## 25  Winter Healesville 4.1616438    8.6  NA                NA        NA
## 26  Winter     Channel 2.2849315    9.5  NA                NA        NA
## 27  Summer     Channel 2.9315068    9.7  NA                NA        NA
## 34  Summer     Channel 2.9150685    9.3  NA                NA        NA
## 36  Autumn Healesville 2.9945205    8.8  NA                NA        NA
## 37  Summer Healesville 1.5890411   4.76  NA                NA        NA
## 39  Summer     Channel 0.8657534    4.9  NA                NA        NA
## 41  Autumn     Channel 3.1013699    8.9  NA                NA        NA
## 43  Summer     Channel 2.9150685    8.8  NA                NA        NA
## 44  Winter     Channel 3.2821918   11.1  NA                NA        NA
## 46  Summer     Channel 1.8712329    7.1  NA                NA        NA
## 50  Winter     Channel 1.2794521    6.3  NA                NA        NA
## 53  Spring     Channel 1.6328767    5.8  NA                NA        NA
## 54  Spring Healesville 1.6493151   8.06  NA                NA        NA
## 55  Summer Healesville 5.6000000   9.02  NA                NA        NA
## 58  Autumn     Channel 3.1726027   11.5  NA                NA        NA
## 60  Autumn     Monarto 2.0767123   7.95  NA                NA        NA
## 61  Winter Healesville 1.2849315   5.58  NA                NA        NA
## 63  Winter     Channel 1.3917808    6.8  NA                NA        NA
## 65  Autumn     Channel 2.1972603    7.7  NA                NA        NA
## 67  Winter     Channel 3.2794521      9  NA                NA        NA
## 68  Autumn     Monarto 2.1616438    N/A  NA                NA        NA
## 69  Winter     Monarto 4.4547945    N/A  NA                NA        NA
## 70  Summer     Monarto 2.5397260   7.94  NA                NA        NA
## 73  Summer     Channel 1.8794521    7.7  NA                NA        NA
## 74  Autumn     Channel 2.1150685    6.8  NA                NA        NA
## 75  Autumn     Monarto 2.0767123    N/A  NA                NA        NA
## 78  Summer Healesville 2.8630137   6.46  NA                NA        NA
## 79  Summer     Channel 1.9178082    4.7  NA                NA        NA
## 80  Spring     Channel 3.6191781    7.4  NA                NA        NA
## 81  Summer Healesville 4.7260274   5.85  NA                NA        NA
## 82  Summer Healesville 3.7863014    4.8  NA                NA        NA
## 83  Spring Healesville 4.5232877   7.88  NA                NA        NA
## 84  Summer     Channel 2.8739726    N/A  NA                NA        NA
## 85  Summer     Channel 2.9287671    8.7  NA                NA        NA
## 86  Summer     Channel 1.9178082    6.8  NA                NA        NA
## 87  Summer     Monarto 4.7726027   7.51  NA                NA        NA
## 88  Summer Healesville 2.7671233   5.82  NA                NA        NA
## 89  Summer     Monarto 4.5589041   8.59  NA                NA        NA
## 90  Summer     Channel 5.8986301    9.8  NA                NA        NA
## 91  Summer     Channel 0.8767123    3.7  NA                NA        NA
## 92  Summer     Channel 1.8821918    7.6  NA                NA        NA
## 93  Winter Healesville 1.3424658   5.42  NA                NA        NA
## 94  Summer     Monarto 1.7917808   6.67  NA                NA        NA
## 95  Winter     Channel 1.3643836    5.1  NA                NA        NA
## 96  Summer Healesville 0.7534247   2.72  NA                NA        NA
## 97  Winter     Channel 1.2684932      6  NA                NA        NA
## 98  Winter     Monarto 1.5150685    4.6  NA                NA        NA
## 99  Winter Healesville 1.4000000   5.82  NA                NA        NA
## 100 Winter Healesville 1.2986301    5.1  NA                NA        NA
## 101 Winter     Channel 1.4082192    6.2  NA                NA        NA
## 102 Winter     Channel 1.3424658    5.1  NA                NA        NA
## 103 Winter     Channel 1.3671233      6  NA                NA        NA
## 104 Summer     Monarto 4.5917808   9.82  NA                NA        NA
## 105 Winter Healesville 1.1123288   6.02  NA                NA        NA
## 106 Autumn Healesville 1.0164384    6.1  NA                NA        NA
## 107 Summer     Channel 1.8849315      9  NA                NA        NA
## 108 Autumn     Channel 1.1945205    6.9  NA                NA        NA
## 109 Summer Healesville 1.4219178   6.62  NA                NA        NA
## 110 Summer Healesville 1.7315068   6.26  NA                NA        NA
## 111 Summer Healesville 1.7479452   7.42  NA                NA        NA
## 112 Summer     Channel 0.9178082    4.1  NA                NA        NA
## 113 Summer     Channel 0.9068493      6  NA                NA        NA
## 114 Summer     Channel 0.9068493    6.2  NA                NA        NA
## 115 Winter     Channel 2.2739726    8.5  NA                NA        NA
## 116 Winter     Monarto 1.6986301   8.75  NA                NA        NA
## 117 Winter Healesville 1.3041096  4.036  NA                NA        NA
## 118 Winter Healesville 1.3616438    6.4  NA                NA        NA
## 119 Winter     Channel 1.3972603    7.2  NA                NA        NA
## 120 Winter     Channel 1.3452055    8.2  NA                NA        NA
## 121 Winter     Monarto 1.5178082   8.15  NA                NA        NA
## 122 Winter Healesville 1.2630137   6.32  NA                NA        NA
## 123 Winter     Channel 1.4438356    8.2  NA                NA        NA
## 124 Winter Healesville 1.2986301   5.76  NA                NA        NA
## 125 Winter     Channel 1.4027397      6  NA                NA        NA
## 126 Winter     Channel 1.2712329    7.6  NA                NA        NA
## 127 Winter     Channel 1.4109589    7.3  NA                NA        NA
## 128 Winter Healesville 1.1753425   5.34  NA                NA        NA
## 129 Winter     Monarto 1.4657534   4.34  NA                NA        NA
## 130 Winter Healesville 1.3835616    N/A  NA                NA        NA
## 131 Winter     Channel 1.2164384    7.3  NA                NA        NA
## 132 Winter     Channel 1.3013699    6.4  NA                NA        NA
## 133 Autumn     Channel 3.1452055    8.3  NA                NA        NA
## 134 Winter     Channel 4.2876712   10.4  NA                NA        NA
## 135 Winter Healesville 5.1041096   8.02  NA                NA        NA
## 136 Autumn Healesville 1.8246575   5.94  NA                NA        NA
## 137 Autumn     Channel 1.1013699    6.6  NA                NA        NA
## 139 Summer Healesville 2.7452055    N/A  NA                NA        NA
## 140 Spring Healesville 6.6273973   7.52  NA                NA        NA
## 141 Winter Healesville 1.1479452   6.54  NA                NA        NA
## 142 Summer Healesville 3.6904110   7.74  NA                NA        NA
## 143 Winter Healesville 4.3013699   9.16  NA                NA        NA
## 144 Spring Healesville 1.6767123   5.86  NA                NA        NA
##     UlcerationDepth Secondary.infection Total.tumour.volume  Barcode.sequence
## 1                                    NA                  NA AGAGTAGC+AAGGCGTA
## 5                                    NA                  NA GAAGGAAG+GACGTCAT
## 7                                    NA                  NA GCAAGATC+AGTCGAAG
## 8                                    NA                  NA GCTATCCT+AACAGGTG
## 9                                    NA                  NA ACGTTCAG+AGTTGTGC
## 10                                   NA                  NA CGCATGAT+AAGCCTGA
## 15                                   NA                  NA TGAGCTAG+GAACGGTT
## 16                                   NA                  NA CCGTAAGA+AGAGTCCA
## 17                                   NA                  NA CAACACCT+AGCTACCA
## 24                                   NA                  NA GATTACCG+TTGGACTG
## 25                                   NA                  NA CACTAGCT+TCGGATTC
## 26                                   NA                  NA CTGTTGAC+ACCTCAGT
## 27                                   NA                  NA CGCAATCT+ACAGGCAT
## 34                                   NA                  NA AGAACGAG+ATTAGCCG
## 36                                   NA                  NA AGTGGATC+TCGATGAC
## 37                                   NA                  NA ATCACACG+GTAGCGTA
## 39                                   NA                  NA TTCCAAGG+CTACAAGG
## 41                                   NA                  NA GCGTCATT+CAGACGTT
## 43                                   NA                  NA CGACGTTA+CTCTGGAT
## 44                                   NA                  NA ATCGATCG+TGGAAGCA
## 46                                   NA                  NA CAGTCCAA+CGGTAATC
## 50                                   NA                  NA AACTGAGC+CAATCAGG
## 53                                   NA                  NA CTTAGGAC+ACTCCTAC
## 54                                   NA                  NA AACGTCTG+AATGACGC
## 55                                   NA                  NA GTGCCATA+CTCCTAGT
## 58                                   NA                  NA CTGATCGT+GCGCATAT
## 60                                   NA                  NA ATCGCCAT+TGCGTAAC
## 61                                   NA                  NA GAACATCG+TCGACAAG
## 63                                   NA                  NA TCGCTGTT+CAAGTCGT
## 65                                   NA                  NA ATGACGTC+CTTCCTTC
## 67                                   NA                  NA TCAGGCTT+ATCATGCG
## 68                                   NA                  NA ACGACTTG+AACAGCGA
## 69                                   NA                  NA CCTTGTAG+CCTTGGAA
## 70                                   NA                  NA GAGCAGTA+TGAGCTGT
## 73                                   NA                  NA ACCACGAT+GTCTGCAA
## 74                                   NA                  NA GAGACGAT+ACCGGTTA
## 75                                   NA                  NA GATAGGCT+CAGTCACA
## 78                                   NA                  NA TGAAGACG+ACATGCCA
## 79                                   NA                  NA ATGACGTC+CTTCCTTC
## 80                                   NA                  NA CATGGCTA+CACACATC
## 81                                   NA                  NA AACGTCTG+AATGACGC
## 82                                   NA                  NA GTCATCGA+GATCCACT
## 83                                   NA                  NA AGCGTGTT+CTTCGCAA
## 84                                   NA                  NA AGAGTAGC+AAGGCGTA
## 85                                   NA                  NA CGCTCTAT+TTGCAACG
## 86                                   NA                  NA AGTTCGTC+ACTCGATC
## 87                                   NA                  NA GCAAGATC+AGTCGAAG
## 88                                   NA                  NA TTGCGAAG+AACACGCT
## 89                                   NA                  NA TGCTTCCA+CGATCGAT
## 90                                   NA                  NA ATCCAGAG+TAACGTCG
## 91                                   NA                  NA AACTGAGC+CAATCAGG
## 92                                   NA                  NA GTCGAAGA+CCACATTG
## 93                                   NA                  NA TGAGCTAG+GAACGGTT
## 94                                   NA                  NA CTTAGGAC+ACTCCTAC
## 95                                   NA                  NA CAATGTGG+TCTTCGAC
## 96                                   NA                  NA TGGTACAG+TCGAGAGT
## 97                                   NA                  NA CGGTTGTT+GTGGTATG
## 98                                   NA                  NA ACCACGAT+GTCTGCAA
## 99                                   NA                  NA ACTGAGGT+GTCAACAG
## 100                                  NA                  NA ACGACTTG+AACAGCGA
## 101                                  NA                  NA GATCGAGT+GACGAACT
## 102                                  NA                  NA CCTGATTG+GCTCAGTT
## 103                                  NA                  NA ATCGATCG+TGGAAGCA
## 104                                  NA                  NA TCAGGCTT+ATCATGCG
## 105                                  NA                  NA ATGCCTGT+AGATTGCG
## 106                                  NA                  NA CCGTAAGA+AGAGTCCA
## 107                                  NA                  NA CACCTGTT+AGGATAGC
## 108                                  NA                  NA TGGACTCT+TCTTACGG
## 109                                  NA                  NA TTCCAAGG+CTACAAGG
## 110                                  NA                  NA GTAGGAGT+GTCCTAAG
## 111                                  NA                  NA CTTCGACT+GATCTTGC
## 112                                  NA                  NA AAGCACTG+AGGTCAAC
## 113                                  NA                  NA CGTTGCAA+ATAGAGCG
## 114                                  NA                  NA GAAGGAAG+GACGTCAT
## 115                                  NA                  NA GTTGACCT+CAGTGCTT
## 116                                  NA                  NA TTGGACGT+GAAGTGCT
## 117                                  NA                  NA TATCGGTC+TACCGGAT
## 118                                  NA                  NA TACGCCTT+GCTACTCT
## 119                                  NA                  NA TTCGTTGG+TACACACG
## 120                                  NA                  NA GATGTGTG+TAGCCATG
## 121                                  NA                  NA CGTGTGTA+CCAACGAA
## 122                                  NA                  NA CCTTGTAG+CCTTGGAA
## 123                                  NA                  NA ACTAGGAG+TATGGCAC
## 124                                  NA                  NA CGCAATCT+ACAGGCAT
## 125                                  NA                  NA ACGGAACA+ACGAGAAC
## 126                                  NA                  NA ACAGCTCA+TACTGCTC
## 127                                  NA                  NA ACTCTCGA+CTGTACCA
## 128                                  NA                  NA AGTGGATC+TCGATGAC
## 129                                  NA                  NA CTGTTGAC+ACCTCAGT
## 130                                  NA                  NA AACCGTTC+CTAGCTCA
## 131                                  NA                  NA GAAGTTGG+CGAACAAC
## 132                                  NA                  NA GATTGCTC+CACTTCAC
## 133                                  NA                  NA CGACGTTA+CTCTGGAT
## 134                                  NA                  NA TGGTAGCT+AGGTGTTG
## 135                                  NA                  NA TAACCGGT+ATCGTCTC
## 136                                  NA                  NA TTGCAGAC+ATCGTGGT
## 137                                  NA                  NA CGCATGAT+AAGCCTGA
## 139                                  NA                  NA GAGACGAT+ACCGGTTA
## 140                                  NA                  NA CTTGTCGA+CGATGTTC
## 141                                  NA                  NA TACGCTAC+CGTGTGAT
## 142                                  NA                  NA GAATCCGA+AGCTAGTG
## 143                                  NA                  NA GCACAACT+CTGAACGT
## 144                                  NA                  NA TGTGACTG+AGCCTATC
##     PF.Clusters X..of.the.lane X..Perfect.barcode X..One.mismatch.barcode
## 1     4,511,702           1.00              98.29                    1.71
## 5     6,479,047           1.00              97.28                    2.72
## 7     3,296,547           1.00              97.40                    2.60
## 8     5,338,890           1.00              98.08                    1.92
## 9    14,968,459           3.00              97.82                    2.18
## 10    4,556,781           1.00              95.70                    4.30
## 15    4,579,681           1.00              96.67                    3.33
## 16    4,340,498           1.00              97.95                    2.05
## 17   16,944,152           3.00              97.98                    2.02
## 24    5,125,566           1.00              96.21                    3.79
## 25    5,317,767           1.00              97.33                    2.67
## 26    5,617,260           1.00              97.81                    2.19
## 27    2,937,660           1.00              97.71                    2.29
## 34    3,120,448           1.00              95.32                    4.68
## 36    2,690,632           1.00              97.77                    2.23
## 37   12,134,750           2.00              96.94                    3.06
## 39    1,800,468           0.00              97.79                    2.21
## 41    6,731,177           1.00              98.04                    1.96
## 43    7,901,958           2.00              98.02                    1.98
## 44    5,257,820           1.00              95.13                    4.87
## 46      454,240           0.00              97.95                    2.05
## 50    1,709,078           0.00              98.04                    1.96
## 53    3,945,235           1.00              98.11                    1.89
## 54    3,809,607           1.00              96.73                    3.27
## 55    9,606,720           2.00              97.52                    2.48
## 58   10,923,019           2.00              98.07                    1.93
## 60    6,018,884           1.00              97.88                    2.12
## 61    6,861,544           1.00              97.76                    2.24
## 63    5,076,673           1.00              97.41                    2.59
## 65    6,756,342           1.00              97.12                    2.88
## 67    3,071,283           1.00              97.28                    2.72
## 68    3,000,501           1.00              97.96                    2.04
## 69    5,927,942           1.00              97.40                    2.60
## 70    1,702,608           0.00              98.07                    1.93
## 73    1,732,316           0.00              97.57                    2.43
## 74        1,152           0.00              96.70                    3.30
## 75    7,628,257           2.00              98.34                    1.66
## 78   13,381,058           3.00              96.98                    3.02
## 79    1,308,213           0.28              97.48                    2.52
## 80    1,840,822           0.40              97.83                    2.17
## 81    8,041,818           1.73              96.32                    3.68
## 82    5,758,111           1.24              96.72                    3.28
## 83    5,774,637           1.24              96.76                    3.24
## 84    7,216,024           1.56              97.99                    2.01
## 85    2,918,728           0.63              97.60                    2.40
## 86   10,758,280           2.32              97.92                    2.08
## 87      348,621           0.08              97.24                    2.76
## 88    6,376,997           1.37              97.60                    2.40
## 89    9,181,668           1.98              98.08                    1.92
## 90    7,154,214           1.54              73.27                   26.73
## 91    7,329,322           1.58              97.89                    2.11
## 92    5,503,188           1.19              97.18                    2.82
## 93    3,740,878           0.81              97.15                    2.85
## 94   11,133,566           2.40              97.85                    2.15
## 95    4,353,832           0.94              97.73                    2.27
## 96    7,887,375           1.70              97.57                    2.43
## 97   11,961,885           2.58              95.85                    4.15
## 98    7,308,907           1.58              97.70                    2.30
## 99    2,984,930           0.64              97.59                    2.41
## 100          98           0.00              94.90                    5.10
## 101   6,506,408           1.40              97.63                    2.37
## 102  11,494,825           2.48              97.55                    2.45
## 103   6,018,258           1.30              92.40                    7.60
## 104   4,879,629           1.05              97.38                    2.62
## 105   4,857,395           1.05              97.58                    2.42
## 106   1,074,345           0.23              97.56                    2.44
## 107           0           0.00                 NA                      NA
## 108  14,104,602           3.04              98.15                    1.85
## 109  10,669,237           2.30              97.64                    2.36
## 110  10,603,179           2.29              97.34                    2.66
## 111   5,642,025           1.22              96.37                    3.63
## 112   4,225,812           0.91              97.88                    2.12
## 113   6,385,393           1.38              95.82                    4.18
## 114   5,390,453           1.16              97.29                    2.71
## 115   5,450,970           1.18              97.62                    2.38
## 116  10,821,895           2.33              96.02                    3.98
## 117   8,693,920           1.87              98.13                    1.87
## 118   3,422,010           0.74              97.72                    2.28
## 119  10,720,172           2.31              97.18                    2.82
## 120   6,051,140           1.30              97.63                    2.37
## 121   4,811,453           1.04              97.67                    2.33
## 122   5,913,586           1.27              97.05                    2.95
## 123   7,229,403           1.56              94.10                    5.90
## 124   8,306,537           1.79              96.94                    3.06
## 125   7,783,546           1.68              97.13                    2.87
## 126   9,742,792           2.10              97.63                    2.37
## 127   6,770,476           1.46              97.75                    2.25
## 128   3,528,471           0.76              97.24                    2.76
## 129          23           0.00             100.00                     NaN
## 130   7,247,793           1.56              97.63                    2.37
## 131   7,212,495           1.55              97.63                    2.37
## 132   4,756,766           1.03              97.26                    2.74
## 133   1,476,427           0.32              97.63                    2.37
## 134   2,413,884           0.52              96.34                    3.66
## 135   6,908,416           1.49              97.09                    2.91
## 136   4,475,461           0.96              95.05                    4.95
## 137  17,006,305           3.67              96.93                    3.07
## 139  12,024,421           2.59              95.37                    4.63
## 140  10,751,797           2.32              97.49                    2.51
## 141  13,015,081           2.81              97.36                    2.64
## 142   3,336,983           0.72              97.10                    2.90
## 143   6,826,313           1.47              97.66                    2.34
## 144   4,566,180           0.98              97.68                    2.32
##     Yield..Mbases. X..PF.Cluster X.....Q30 Mean.Quality X.2 X.3 X.4
## 1            1,363           100     93.82        35.92  NA  NA  NA
## 5            1,957           100     93.85        35.94  NA  NA  NA
## 7              996           100     94.12        35.98  NA  NA  NA
## 8            1,612           100     93.69        35.91  NA  NA  NA
## 9            4,520           100     93.68        35.91  NA  NA  NA
## 10           1,376           100     93.94        35.94  NA  NA  NA
## 15           1,383           100     93.75        35.92  NA  NA  NA
## 16           1,311           100     93.54        35.88  NA  NA  NA
## 17           5,117           100     93.05        35.78  NA  NA  NA
## 24           1,548           100     93.89        35.94  NA  NA  NA
## 25           1,606           100     93.24        35.83  NA  NA  NA
## 26           1,696           100     93.55        35.88  NA  NA  NA
## 27             887           100     93.75        35.91  NA  NA  NA
## 34             942           100     94.24        36.01  NA  NA  NA
## 36             813           100     93.77        35.93  NA  NA  NA
## 37           3,665           100     93.34        35.84  NA  NA  NA
## 39             544           100     93.58        35.88  NA  NA  NA
## 41           2,033           100     93.62        35.89  NA  NA  NA
## 43           2,386           100     94.09        35.98  NA  NA  NA
## 44           1,588           100     93.81        35.93  NA  NA  NA
## 46             137           100     92.97        35.76  NA  NA  NA
## 50             516           100     93.06        35.78  NA  NA  NA
## 53           1,191           100     93.55        35.88  NA  NA  NA
## 54           1,151           100     93.07        35.79  NA  NA  NA
## 55           2,901           100     93.25        35.83  NA  NA  NA
## 58           3,299           100     93.51        35.87  NA  NA  NA
## 60           1,818           100     93.58        35.89  NA  NA  NA
## 61           2,072           100     93.49        35.87  NA  NA  NA
## 63           1,533           100     93.77        35.92  NA  NA  NA
## 65           2,040           100     92.53        35.68  NA  NA  NA
## 67             928           100     92.94        35.76  NA  NA  NA
## 68             906           100     93.70        35.91  NA  NA  NA
## 69           1,790           100     93.86        35.94  NA  NA  NA
## 70             514           100     93.58        35.89  NA  NA  NA
## 73             523           100     93.82        35.92  NA  NA  NA
## 74               0           100     92.06        35.56  NA  NA  NA
## 75           2,304           100     93.62        35.89  NA  NA  NA
## 78           4,041           100     93.36        35.84  NA  NA  NA
## 79             395           100     93.03        35.79  NA  NA  NA
## 80             556           100     92.66        35.71  NA  NA  NA
## 81           2,429           100     92.18        35.63  NA  NA  NA
## 82           1,739           100     91.87        35.55  NA  NA  NA
## 83           1,744           100     91.56        35.50  NA  NA  NA
## 84           2,179           100     93.47        35.87  NA  NA  NA
## 85             881           100     93.65        35.91  NA  NA  NA
## 86           3,249           100     93.20        35.82  NA  NA  NA
## 87             105           100     92.77        35.73  NA  NA  NA
## 88           1,926           100     92.44        35.67  NA  NA  NA
## 89           2,773           100     93.04        35.79  NA  NA  NA
## 90           2,161           100     92.95        35.78  NA  NA  NA
## 91           2,213           100     92.63        35.71  NA  NA  NA
## 92           1,662           100     92.53        35.69  NA  NA  NA
## 93           1,130           100     93.03        35.79  NA  NA  NA
## 94           3,362           100     92.47        35.68  NA  NA  NA
## 95           1,315           100     93.72        35.92  NA  NA  NA
## 96           2,382           100     93.15        35.81  NA  NA  NA
## 97           3,612           100     92.97        35.78  NA  NA  NA
## 98           2,207           100     92.84        35.75  NA  NA  NA
## 99             901           100     92.49        35.69  NA  NA  NA
## 100              0           100     89.11        34.91  NA  NA  NA
## 101          1,965           100     93.11        35.79  NA  NA  NA
## 102          3,471           100     93.66        35.92  NA  NA  NA
## 103          1,818           100     93.20        35.82  NA  NA  NA
## 104          1,474           100     92.63        35.71  NA  NA  NA
## 105          1,467           100     92.64        35.72  NA  NA  NA
## 106            324           100     92.66        35.72  NA  NA  NA
## 107              0            NA        NA           NA  NA  NA  NA
## 108          4,260           100     93.26        35.83  NA  NA  NA
## 109          3,222           100     92.79        35.74  NA  NA  NA
## 110          3,202           100     92.01        35.58  NA  NA  NA
## 111          1,704           100     91.87        35.56  NA  NA  NA
## 112          1,276           100     93.16        35.81  NA  NA  NA
## 113          1,928           100     92.04        35.61  NA  NA  NA
## 114          1,628           100     93.28        35.84  NA  NA  NA
## 115          1,646           100     93.04        35.79  NA  NA  NA
## 116          3,268           100     92.72        35.73  NA  NA  NA
## 117          2,626           100     92.32        35.65  NA  NA  NA
## 118          1,033           100     92.94        35.76  NA  NA  NA
## 119          3,237           100     92.73        35.72  NA  NA  NA
## 120          1,827           100     93.21        35.83  NA  NA  NA
## 121          1,453           100     92.30        35.65  NA  NA  NA
## 122          1,786           100     92.92        35.77  NA  NA  NA
## 123          2,183           100     93.10        35.80  NA  NA  NA
## 124          2,509           100     91.86        35.55  NA  NA  NA
## 125          2,351           100     92.96        35.78  NA  NA  NA
## 126          2,942           100     93.00        35.78  NA  NA  NA
## 127          2,045           100     93.04        35.79  NA  NA  NA
## 128          1,066           100     93.08        35.79  NA  NA  NA
## 129              0           100     95.85        36.37  NA  NA  NA
## 130          2,189           100     92.28        35.63  NA  NA  NA
## 131          2,178           100     93.09        35.80  NA  NA  NA
## 132          1,437           100     93.49        35.88  NA  NA  NA
## 133            446           100     93.25        35.82  NA  NA  NA
## 134            729           100     92.34        35.64  NA  NA  NA
## 135          2,086           100     91.90        35.56  NA  NA  NA
## 136          1,352           100     91.42        35.46  NA  NA  NA
## 137          5,136           100     92.91        35.77  NA  NA  NA
## 139          3,631           100     93.06        35.79  NA  NA  NA
## 140          3,247           100     92.77        35.74  NA  NA  NA
## 141          3,931           100     92.61        35.70  NA  NA  NA
## 142          1,008           100     93.09        35.79  NA  NA  NA
## 143          2,062           100     92.67        35.71  NA  NA  NA
## 144          1,379           100     92.13        35.61  NA  NA  NA

ssWvCAV <- ssWvCAV[ssWvCAV$Sample %in% colnames(x),]

dim(ssWvCAV)

## [1] 96 31

x1 <- x[,which(colnames(x) %in% ssWvCAV$Sample)]
dim(x1)

## [1] 310  96

x1 <- x1[which(rowMeans(x1)>10),]
dim(x1)

## [1] 261  96

ssWvCAV %>%  kbl(caption="Sample sheet")%>% kable_styling("hover",full_width=FALSE) 

Sample sheet

	Lane	Project	Client.ID	Sample	X	Category	Sex.Age.Season	Sex	Age	Season	Location	Agebyyear	Weight	X.1	Number.of.Tumours	Ulcerated	UlcerationDepth	Secondary.infection	Total.tumour.volume	Barcode.sequence	PF.Clusters	X..of.the.lane	X..Perfect.barcode	X..One.mismatch.barcode	Yield..Mbases.	X..PF.Cluster	X…..Q30	Mean.Quality	X.2	X.3	X.4
1	1	default	S1_LauraGrace	6102461	NA	Healthy/Wild	FAP	F	A	Spring	Channel	2.6219178	7.2	NA	NA	NA		NA	NA	AGAGTAGC+AAGGCGTA	4,511,702	1.00	98.29	1.71	1,363	100	93.82	35.92	NA	NA	NA
5	1	default	S1_Turmeric	6102465	NA	Healthy/Wild	FAU	F	A	Summer	Channel	1.9095890	7.8	NA	NA	NA		NA	NA	GAAGGAAG+GACGTCAT	6,479,047	1.00	97.28	2.72	1,957	100	93.85	35.94	NA	NA	NA
7	1	default	S2_Belinda	6102467	NA	Healthy/Wild	FAU	F	A	Summer	Channel	1.8657534	6.6	NA	NA	NA		NA	NA	GCAAGATC+AGTCGAAG	3,296,547	1.00	97.40	2.60	996	100	94.12	35.98	NA	NA	NA
8	1	default	S2_B99008	6102468	NA	Captive	FAU	F	A	Summer	Monarto	3.7369863	8.12	NA	NA	NA		NA	NA	GCTATCCT+AACAGGTG	5,338,890	1.00	98.08	1.92	1,612	100	93.69	35.91	NA	NA	NA
9	1	default	S2_Bonney	6102469	NA	Captive	FAU	F	A	Summer	Healesville	2.5863014	5.4	NA	NA	NA		NA	NA	ACGTTCAG+AGTTGTGC	14,968,459	3.00	97.82	2.18	4,520	100	93.68	35.91	NA	NA	NA
10	1	default	S2_Rosie	6102470	NA	Healthy/Wild	FAU	F	A	Summer	Channel	1.8904110	7.4	NA	NA	NA		NA	NA	CGCATGAT+AAGCCTGA	4,556,781	1.00	95.70	4.30	1,376	100	93.94	35.94	NA	NA	NA
15	1	default	S3_ST62	6102475	NA	Healthy/Wild	FAW	F	A	Winter	Channel	3.2712329	10.2	NA	NA	NA		NA	NA	TGAGCTAG+GAACGGTT	4,579,681	1.00	96.67	3.33	1,383	100	93.75	35.92	NA	NA	NA
16	1	default	S3_Jessie	6102476	NA	Captive	FAW	F	A	Winter	Monarto	2.1945205	12.95	NA	NA	NA		NA	NA	CCGTAAGA+AGAGTCCA	4,340,498	1.00	97.95	2.05	1,311	100	93.54	35.88	NA	NA	NA
17	1	default	S3_McCauley	6102477	NA	Captive	MAA	M	A	Autumn	Healesville	3.0328767	8.52	NA	NA	NA		NA	NA	CAACACCT+AGCTACCA	16,944,152	3.00	97.98	2.02	5,117	100	93.05	35.78	NA	NA	NA
24	1	default	S5_Fenton	6102484	NA	Captive	MAU	M	A	Summer	Monarto	3.0000000	N/A	NA	NA	NA		NA	NA	GATTACCG+TTGGACTG	5,125,566	1.00	96.21	3.79	1,548	100	93.89	35.94	NA	NA	NA
25	1	default	S5_Bryan Adams	6102485	NA	Captive	MAW	M	A	Winter	Healesville	4.1616438	8.6	NA	NA	NA		NA	NA	CACTAGCT+TCGGATTC	5,317,767	1.00	97.33	2.67	1,606	100	93.24	35.83	NA	NA	NA
26	1	default	S5_Silky	6102486	NA	Healthy/Wild	MAW	M	A	Winter	Channel	2.2849315	9.5	NA	NA	NA		NA	NA	CTGTTGAC+ACCTCAGT	5,617,260	1.00	97.81	2.19	1,696	100	93.55	35.88	NA	NA	NA
27	1	default	S5_Longley	6102487	NA	Healthy/Wild	MAU	M	A	Summer	Channel	2.9315068	9.7	NA	NA	NA		NA	NA	CGCAATCT+ACAGGCAT	2,937,660	1.00	97.71	2.29	887	100	93.75	35.91	NA	NA	NA
34	1	default	S6_Ben	6102494	NA	Wild	MAU	M	A	Summer	Channel	2.9150685	9.3	NA	NA	NA		NA	NA	AGAACGAG+ATTAGCCG	3,120,448	1.00	95.32	4.68	942	100	94.24	36.01	NA	NA	NA
36	1	default	S7_Fredo	6102496	NA	Captive	MAA	M	A	Autumn	Healesville	2.9945205	8.8	NA	NA	NA		NA	NA	AGTGGATC+TCGATGAC	2,690,632	1.00	97.77	2.23	813	100	93.77	35.93	NA	NA	NA
37	1	default	S7_Piglet	6102497	NA	Captive	FSU	F	S	Summer	Healesville	1.5890411	4.76	NA	NA	NA		NA	NA	ATCACACG+GTAGCGTA	12,134,750	2.00	96.94	3.06	3,665	100	93.34	35.84	NA	NA	NA
39	1	default	S7_Lieke	6102499	NA	Healthy/Wild	FSU	F	S	Summer	Channel	0.8657534	4.9	NA	NA	NA		NA	NA	TTCCAAGG+CTACAAGG	1,800,468	0.00	97.79	2.21	544	100	93.58	35.88	NA	NA	NA
41	1	default	S7_Smithers	6102501	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	3.1013699	8.9	NA	NA	NA		NA	NA	GCGTCATT+CAGACGTT	6,731,177	1.00	98.04	1.96	2,033	100	93.62	35.89	NA	NA	NA
43	1	default	S8_Vinnie	6102503	NA	Healthy/Wild	MAU	M	A	Summer	Channel	2.9150685	8.8	NA	NA	NA		NA	NA	CGACGTTA+CTCTGGAT	7,901,958	2.00	98.02	1.98	2,386	100	94.09	35.98	NA	NA	NA
44	1	default	S8_Struggles	6102504	NA	Healthy/Wild	MAW	M	A	Winter	Channel	3.2821918	11.1	NA	NA	NA		NA	NA	ATCGATCG+TGGAAGCA	5,257,820	1.00	95.13	4.87	1,588	100	93.81	35.93	NA	NA	NA
50	1	default	S9_Snoopy	6102510	NA	Healthy/Wild	MSW	M	S	Winter	Channel	1.2794521	6.3	NA	NA	NA		NA	NA	AACTGAGC+CAATCAGG	1,709,078	0.00	98.04	1.96	516	100	93.06	35.78	NA	NA	NA
53	1	default	S9_Sarabi	6102513	NA	Healthy/Wild	FSP	F	S	Spring	Channel	1.6328767	5.8	NA	NA	NA		NA	NA	CTTAGGAC+ACTCCTAC	3,945,235	1.00	98.11	1.89	1,191	100	93.55	35.88	NA	NA	NA
54	1	default	S10_McMahon	6102514	NA	Captive	MSP	M	S	Spring	Healesville	1.6493151	8.06	NA	NA	NA		NA	NA	AACGTCTG+AATGACGC	3,809,607	1.00	96.73	3.27	1,151	100	93.07	35.79	NA	NA	NA
55	1	default	S10_Logan	6102515	NA	Captive	MAU	M	A	Summer	Healesville	5.6000000	9.02	NA	NA	NA		NA	NA	GTGCCATA+CTCCTAGT	9,606,720	2.00	97.52	2.48	2,901	100	93.25	35.83	NA	NA	NA
58	1	default	S11_Baloo	6102518	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	3.1726027	11.5	NA	NA	NA		NA	NA	CTGATCGT+GCGCATAT	10,923,019	2.00	98.07	1.93	3,299	100	93.51	35.87	NA	NA	NA
60	1	default	S11_Xuki	6102520	NA	Captive	MAA	M	A	Autumn	Monarto	2.0767123	7.95	NA	NA	NA		NA	NA	ATCGCCAT+TGCGTAAC	6,018,884	1.00	97.88	2.12	1,818	100	93.58	35.89	NA	NA	NA
61	1	default	S11_Kalamata	6102521	NA	Captive	MSW	M	S	Winter	Healesville	1.2849315	5.58	NA	NA	NA		NA	NA	GAACATCG+TCGACAAG	6,861,544	1.00	97.76	2.24	2,072	100	93.49	35.87	NA	NA	NA
63	1	default	S11_DanRhodes	6102523	NA	Healthy/Wild	MSW	M	S	Winter	Channel	1.3917808	6.8	NA	NA	NA		NA	NA	TCGCTGTT+CAAGTCGT	5,076,673	1.00	97.41	2.59	1,533	100	93.77	35.92	NA	NA	NA
65	1	default	S12_Robin	6102525	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	2.1972603	7.7	NA	NA	NA		NA	NA	ATGACGTC+CTTCCTTC	6,756,342	1.00	97.12	2.88	2,040	100	92.53	35.68	NA	NA	NA
67	1	default	S12_Eeyore	6102527	NA	Healthy/Wild	MAW	M	A	Winter	Channel	3.2794521	9	NA	NA	NA		NA	NA	TCAGGCTT+ATCATGCG	3,071,283	1.00	97.28	2.72	928	100	92.94	35.76	NA	NA	NA
68	1	default	S12_Wally	6102528	NA	Captive	MAA	M	A	Autumn	Monarto	2.1616438	N/A	NA	NA	NA		NA	NA	ACGACTTG+AACAGCGA	3,000,501	1.00	97.96	2.04	906	100	93.70	35.91	NA	NA	NA
69	1	default	S12_Naz	6102529	NA	Captive	MAW	M	A	Winter	Monarto	4.4547945	N/A	NA	NA	NA		NA	NA	CCTTGTAG+CCTTGGAA	5,927,942	1.00	97.40	2.60	1,790	100	93.86	35.94	NA	NA	NA
70	1	default	S13_Humbug	6102530	NA	Captive	MAU	M	A	Summer	Monarto	2.5397260	7.94	NA	NA	NA		NA	NA	GAGCAGTA+TGAGCTGT	1,702,608	0.00	98.07	1.93	514	100	93.58	35.89	NA	NA	NA
73	1	default	S13_Pangea	6102533	NA	Healthy/Wild	MAU	M	A	Summer	Channel	1.8794521	7.7	NA	NA	NA		NA	NA	ACCACGAT+GTCTGCAA	1,732,316	0.00	97.57	2.43	523	100	93.82	35.92	NA	NA	NA
75	1	default	S13_Xerxes	6102535	NA	Captive	MAA	M	A	Autumn	Monarto	2.0767123	N/A	NA	NA	NA		NA	NA	GATAGGCT+CAGTCACA	7,628,257	2.00	98.34	1.66	2,304	100	93.62	35.89	NA	NA	NA
78	1	default	S14_Catani	6102538	NA	Captive	FAU	F	A	Summer	Healesville	2.8630137	6.46	NA	NA	NA		NA	NA	TGAAGACG+ACATGCCA	13,381,058	3.00	96.98	3.02	4,041	100	93.36	35.84	NA	NA	NA
79	2	default	S15_Halls	6102539	NA	Wild	FAU	F	A	Summer	Channel	1.9178082	4.7	NA	NA	NA		NA	NA	ATGACGTC+CTTCCTTC	1,308,213	0.28	97.48	2.52	395	100	93.03	35.79	NA	NA	NA
80	2	default	S15_Jessica	6102540	NA	Wild	FAP	F	A	Spring	Channel	3.6191781	7.4	NA	NA	NA		NA	NA	CATGGCTA+CACACATC	1,840,822	0.40	97.83	2.17	556	100	92.66	35.71	NA	NA	NA
81	2	default	S15_Amelia	6102541	NA	Captive	FAU	F	A	Summer	Healesville	4.7260274	5.85	NA	NA	NA		NA	NA	AACGTCTG+AATGACGC	8,041,818	1.73	96.32	3.68	2,429	100	92.18	35.63	NA	NA	NA
82	2	default	S15_London	6102542	NA	Captive	FAU	F	A	Summer	Healesville	3.7863014	4.8	NA	NA	NA		NA	NA	GTCATCGA+GATCCACT	5,758,111	1.24	96.72	3.28	1,739	100	91.87	35.55	NA	NA	NA
83	2	default	S15_Elyse	6102543	NA	Captive	FAP	F	A	Spring	Healesville	4.5232877	7.88	NA	NA	NA		NA	NA	AGCGTGTT+CTTCGCAA	5,774,637	1.24	96.76	3.24	1,744	100	91.56	35.50	NA	NA	NA
84	2	default	S15_Munchkin	6102544	NA	Wild	FAU	F	A	Summer	Channel	2.8739726	N/A	NA	NA	NA		NA	NA	AGAGTAGC+AAGGCGTA	7,216,024	1.56	97.99	2.01	2,179	100	93.47	35.87	NA	NA	NA
85	2	default	S16_Pelverata	6102545	NA	Wild	FAU	F	A	Summer	Channel	2.9287671	8.7	NA	NA	NA		NA	NA	CGCTCTAT+TTGCAACG	2,918,728	0.63	97.60	2.40	881	100	93.65	35.91	NA	NA	NA
86	2	default	S16_Lizzie	6102546	NA	Wild	FAU	F	A	Summer	Channel	1.9178082	6.8	NA	NA	NA		NA	NA	AGTTCGTC+ACTCGATC	10,758,280	2.32	97.92	2.08	3,249	100	93.20	35.82	NA	NA	NA
88	2	default	S16_Milla	6102548	NA	Captive	FAU	F	A	Summer	Healesville	2.7671233	5.82	NA	NA	NA		NA	NA	TTGCGAAG+AACACGCT	6,376,997	1.37	97.60	2.40	1,926	100	92.44	35.67	NA	NA	NA
89	2	default	S16_Maleficent	6102549	NA	Captive	FAU	F	A	Summer	Monarto	4.5589041	8.59	NA	NA	NA		NA	NA	TGCTTCCA+CGATCGAT	9,181,668	1.98	98.08	1.92	2,773	100	93.04	35.79	NA	NA	NA
90	2	default	S16_Lebanon	6102550	NA	Wild	FAU	F	A	Summer	Channel	5.8986301	9.8	NA	NA	NA		NA	NA	ATCCAGAG+TAACGTCG	7,154,214	1.54	73.27	26.73	2,161	100	92.95	35.78	NA	NA	NA
91	2	default	S17_Radiata	6102551	NA	Wild	FSU	F	S	Summer	Channel	0.8767123	3.7	NA	NA	NA		NA	NA	AACTGAGC+CAATCAGG	7,329,322	1.58	97.89	2.11	2,213	100	92.63	35.71	NA	NA	NA
92	2	default	S17_Shenzi	6102552	NA	Wild	FAU	F	A	Summer	Channel	1.8821918	7.6	NA	NA	NA		NA	NA	GTCGAAGA+CCACATTG	5,503,188	1.19	97.18	2.82	1,662	100	92.53	35.69	NA	NA	NA
93	2	default	S17_Anirak	6102553	NA	Captive	FSW	F	S	Winter	Healesville	1.3424658	5.42	NA	NA	NA		NA	NA	TGAGCTAG+GAACGGTT	3,740,878	0.81	97.15	2.85	1,130	100	93.03	35.79	NA	NA	NA
94	2	default	S17_Poly	6102554	NA	Captive	FSU	F	S	Summer	Monarto	1.7917808	6.67	NA	NA	NA		NA	NA	CTTAGGAC+ACTCCTAC	11,133,566	2.40	97.85	2.15	3,362	100	92.47	35.68	NA	NA	NA
95	2	default	S17_Vernicosa	6102555	NA	Wild	FSW	F	S	Winter	Channel	1.3643836	5.1	NA	NA	NA		NA	NA	CAATGTGG+TCTTCGAC	4,353,832	0.94	97.73	2.27	1,315	100	93.72	35.92	NA	NA	NA
96	2	default	S17_Methyl	6102556	NA	Captive	FSU	F	S	Summer	Healesville	0.7534247	2.72	NA	NA	NA		NA	NA	TGGTACAG+TCGAGAGT	7,887,375	1.70	97.57	2.43	2,382	100	93.15	35.81	NA	NA	NA
97	2	default	S18_Paprika	6102557	NA	Wild	FSW	F	S	Winter	Channel	1.2684932	6	NA	NA	NA		NA	NA	CGGTTGTT+GTGGTATG	11,961,885	2.58	95.85	4.15	3,612	100	92.97	35.78	NA	NA	NA
98	2	default	S18_Zara	6102558	NA	Captive	FSW	F	S	Winter	Monarto	1.5150685	4.6	NA	NA	NA		NA	NA	ACCACGAT+GTCTGCAA	7,308,907	1.58	97.70	2.30	2,207	100	92.84	35.75	NA	NA	NA
99	2	default	S18_Maze	6102559	NA	Captive	FSW	F	S	Winter	Healesville	1.4000000	5.82	NA	NA	NA		NA	NA	ACTGAGGT+GTCAACAG	2,984,930	0.64	97.59	2.41	901	100	92.49	35.69	NA	NA	NA
101	2	default	S18_Haiko	6102561	NA	Wild	FSW	F	S	Winter	Channel	1.4082192	6.2	NA	NA	NA		NA	NA	GATCGAGT+GACGAACT	6,506,408	1.40	97.63	2.37	1,965	100	93.11	35.79	NA	NA	NA
102	2	default	S18_Hermione	6102562	NA	Wild	FSW	F	S	Winter	Channel	1.3424658	5.1	NA	NA	NA		NA	NA	CCTGATTG+GCTCAGTT	11,494,825	2.48	97.55	2.45	3,471	100	93.66	35.92	NA	NA	NA
103	2	default	S19_Tammy	6102563	NA	Wild	FSW	F	S	Winter	Channel	1.3671233	6	NA	NA	NA		NA	NA	ATCGATCG+TGGAAGCA	6,018,258	1.30	92.40	7.60	1,818	100	93.20	35.82	NA	NA	NA
104	2	default	S19_Morocco	6102564	NA	Captive	MAU	M	A	Summer	Monarto	4.5917808	9.82	NA	NA	NA		NA	NA	TCAGGCTT+ATCATGCG	4,879,629	1.05	97.38	2.62	1,474	100	92.63	35.71	NA	NA	NA
105	2	default	S19_Pecorino	6102565	NA	Captive	FSW	F	S	Winter	Healesville	1.1123288	6.02	NA	NA	NA		NA	NA	ATGCCTGT+AGATTGCG	4,857,395	1.05	97.58	2.42	1,467	100	92.64	35.72	NA	NA	NA
108	2	default	S19_Tantor	6102568	NA	Wild	MSA	M	S	Autumn	Channel	1.1945205	6.9	NA	NA	NA		NA	NA	TGGACTCT+TCTTACGG	14,104,602	3.04	98.15	1.85	4,260	100	93.26	35.83	NA	NA	NA
109	2	default	S20_Muasy	6102569	NA	Captive	MSU	M	S	Summer	Healesville	1.4219178	6.62	NA	NA	NA		NA	NA	TTCCAAGG+CTACAAGG	10,669,237	2.30	97.64	2.36	3,222	100	92.79	35.74	NA	NA	NA
110	2	default	S20_Fuse	6102570	NA	Captive	MSU	M	S	Summer	Healesville	1.7315068	6.26	NA	NA	NA		NA	NA	GTAGGAGT+GTCCTAAG	10,603,179	2.29	97.34	2.66	3,202	100	92.01	35.58	NA	NA	NA
111	2	default	S20_Joseph	6102571	NA	Captive	MSU	M	S	Summer	Healesville	1.7479452	7.42	NA	NA	NA		NA	NA	CTTCGACT+GATCTTGC	5,642,025	1.22	96.37	3.63	1,704	100	91.87	35.56	NA	NA	NA
112	2	default	S20_Hudson	6102572	NA	Wild	MSU	M	S	Summer	Channel	0.9178082	4.1	NA	NA	NA		NA	NA	AAGCACTG+AGGTCAAC	4,225,812	0.91	97.88	2.12	1,276	100	93.16	35.81	NA	NA	NA
113	2	default	S20_Mace	6102573	NA	Wild	MSU	M	S	Summer	Channel	0.9068493	6	NA	NA	NA		NA	NA	CGTTGCAA+ATAGAGCG	6,385,393	1.38	95.82	4.18	1,928	100	92.04	35.61	NA	NA	NA
114	2	default	S20_Mint	6102574	NA	Wild	MSU	M	S	Summer	Channel	0.9068493	6.2	NA	NA	NA		NA	NA	GAAGGAAG+GACGTCAT	5,390,453	1.16	97.29	2.71	1,628	100	93.28	35.84	NA	NA	NA
115	2	default	S21_Sylvester	6102575	NA	Wild	MAW	M	A	Winter	Channel	2.2739726	8.5	NA	NA	NA		NA	NA	GTTGACCT+CAGTGCTT	5,450,970	1.18	97.62	2.38	1,646	100	93.04	35.79	NA	NA	NA
116	2	default	S21_Moama	6102576	NA	Captive	MAW	M	A	Winter	Monarto	1.6986301	8.75	NA	NA	NA		NA	NA	TTGGACGT+GAAGTGCT	10,821,895	2.33	96.02	3.98	3,268	100	92.72	35.73	NA	NA	NA
117	2	default	S21_Nauset	6102577	NA	Captive	MSW	M	S	Winter	Healesville	1.3041096	4.036	NA	NA	NA		NA	NA	TATCGGTC+TACCGGAT	8,693,920	1.87	98.13	1.87	2,626	100	92.32	35.65	NA	NA	NA
118	2	default	S21_Graham	6102578	NA	Captive	MSW	M	S	Winter	Healesville	1.3616438	6.4	NA	NA	NA		NA	NA	TACGCCTT+GCTACTCT	3,422,010	0.74	97.72	2.28	1,033	100	92.94	35.76	NA	NA	NA
119	2	default	S21_Harry	6102579	NA	Wild	MSW	M	S	Winter	Channel	1.3972603	7.2	NA	NA	NA		NA	NA	TTCGTTGG+TACACACG	10,720,172	2.31	97.18	2.82	3,237	100	92.73	35.72	NA	NA	NA
120	2	default	S21_Thor	6102580	NA	Wild	MSW	M	S	Winter	Channel	1.3452055	8.2	NA	NA	NA		NA	NA	GATGTGTG+TAGCCATG	6,051,140	1.30	97.63	2.37	1,827	100	93.21	35.83	NA	NA	NA
121	2	default	S22_Yanni	6102581	NA	Captive	MSW	M	S	Winter	Monarto	1.5178082	8.15	NA	NA	NA		NA	NA	CGTGTGTA+CCAACGAA	4,811,453	1.04	97.67	2.33	1,453	100	92.30	35.65	NA	NA	NA
122	2	default	S22_Gherkin	6102582	NA	Captive	MSW	M	S	Winter	Healesville	1.2630137	6.32	NA	NA	NA		NA	NA	CCTTGTAG+CCTTGGAA	5,913,586	1.27	97.05	2.95	1,786	100	92.92	35.77	NA	NA	NA
123	2	default	S22_Utah	6102583	NA	Wild	MSW	M	S	Winter	Channel	1.4438356	8.2	NA	NA	NA		NA	NA	ACTAGGAG+TATGGCAC	7,229,403	1.56	94.10	5.90	2,183	100	93.10	35.80	NA	NA	NA
124	2	default	S22_Stilton	6102584	NA	Captive	MSW	M	S	Winter	Healesville	1.2986301	5.76	NA	NA	NA		NA	NA	CGCAATCT+ACAGGCAT	8,306,537	1.79	96.94	3.06	2,509	100	91.86	35.55	NA	NA	NA
125	2	default	S22_Wellington	6102585	NA	Wild	MSW	M	S	Winter	Channel	1.4027397	6	NA	NA	NA		NA	NA	ACGGAACA+ACGAGAAC	7,783,546	1.68	97.13	2.87	2,351	100	92.96	35.78	NA	NA	NA
126	2	default	S22_Ginger	6102586	NA	Healthy/Wild	MSW	M	S	Winter	Channel	1.2712329	7.6	NA	NA	NA		NA	NA	ACAGCTCA+TACTGCTC	9,742,792	2.10	97.63	2.37	2,942	100	93.00	35.78	NA	NA	NA
127	2	default	S23_Meliza	6102587	NA	Ignore-Wild	FSW	F	S	Winter	Channel	1.4109589	7.3	NA	NA	NA		NA	NA	ACTCTCGA+CTGTACCA	6,770,476	1.46	97.75	2.25	2,045	100	93.04	35.79	NA	NA	NA
128	2	default	S23_Osiris	6102588	NA	Captive	MSW	M	S	Winter	Healesville	1.1753425	5.34	NA	NA	NA		NA	NA	AGTGGATC+TCGATGAC	3,528,471	0.76	97.24	2.76	1,066	100	93.08	35.79	NA	NA	NA
130	2	default	S23_Quince	6102590	NA	Captive	MSW	M	S	Winter	Healesville	1.3835616	N/A	NA	NA	NA		NA	NA	AACCGTTC+CTAGCTCA	7,247,793	1.56	97.63	2.37	2,189	100	92.28	35.63	NA	NA	NA
131	2	default	S23_Saffron	6102591	NA	Wild	MSW	M	S	Winter	Channel	1.2164384	7.3	NA	NA	NA		NA	NA	GAAGTTGG+CGAACAAC	7,212,495	1.55	97.63	2.37	2,178	100	93.09	35.80	NA	NA	NA
132	2	default	S23_Gandalf	6102592	NA	Wild	MSW	M	S	Winter	Channel	1.3013699	6.4	NA	NA	NA		NA	NA	GATTGCTC+CACTTCAC	4,756,766	1.03	97.26	2.74	1,437	100	93.49	35.88	NA	NA	NA
133	2	default	S24_Boy	6102593	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	3.1452055	8.3	NA	NA	NA		NA	NA	CGACGTTA+CTCTGGAT	1,476,427	0.32	97.63	2.37	446	100	93.25	35.82	NA	NA	NA
134	2	default	S24_ST74	6102594	NA	Wild	MAW	M	A	Winter	Channel	4.2876712	10.4	NA	NA	NA		NA	NA	TGGTAGCT+AGGTGTTG	2,413,884	0.52	96.34	3.66	729	100	92.34	35.64	NA	NA	NA
135	2	default	S24_Malfoy	6102595	NA	Captive	MAW	M	A	Winter	Healesville	5.1041096	8.02	NA	NA	NA		NA	NA	TAACCGGT+ATCGTCTC	6,908,416	1.49	97.09	2.91	2,086	100	91.90	35.56	NA	NA	NA
136	2	default	S24_Licorice	6102596	NA	Captive	FSA	F	S	Autumn	Healesville	1.8246575	5.94	NA	NA	NA		NA	NA	TTGCAGAC+ATCGTGGT	4,475,461	0.96	95.05	4.95	1,352	100	91.42	35.46	NA	NA	NA
137	2	default	S24_Lisa	6102597	NA	Wild	FSA	F	S	Autumn	Channel	1.1013699	6.6	NA	NA	NA		NA	NA	CGCATGAT+AAGCCTGA	17,006,305	3.67	96.93	3.07	5,136	100	92.91	35.77	NA	NA	NA
139	2	default	S25_Tofu	6102599	NA	Captive	FAU	F	A	Summer	Healesville	2.7452055	N/A	NA	NA	NA		NA	NA	GAGACGAT+ACCGGTTA	12,024,421	2.59	95.37	4.63	3,631	100	93.06	35.79	NA	NA	NA
140	2	default	S25_Colette	6102600	NA	Captive	FAP	F	A	Spring	Healesville	6.6273973	7.52	NA	NA	NA		NA	NA	CTTGTCGA+CGATGTTC	10,751,797	2.32	97.49	2.51	3,247	100	92.77	35.74	NA	NA	NA
141	2	default	S25_CannonBall	6102601	NA	Captive	MSW	M	S	Winter	Healesville	1.1479452	6.54	NA	NA	NA		NA	NA	TACGCTAC+CGTGTGAT	13,015,081	2.81	97.36	2.64	3,931	100	92.61	35.70	NA	NA	NA
142	2	default	S25_Cornelius	6102602	NA	Captive	MAU	M	A	Summer	Healesville	3.6904110	7.74	NA	NA	NA		NA	NA	GAATCCGA+AGCTAGTG	3,336,983	0.72	97.10	2.90	1,008	100	93.09	35.79	NA	NA	NA
143	2	default	S25_Mardook	6102603	NA	Captive	MAW	M	A	Winter	Healesville	4.3013699	9.16	NA	NA	NA		NA	NA	GCACAACT+CTGAACGT	6,826,313	1.47	97.66	2.34	2,062	100	92.67	35.71	NA	NA	NA
144	2	default	S25_Negroni	6102604	NA	Captive	FSP	F	S	Spring	Healesville	1.6767123	5.86	NA	NA	NA		NA	NA	TGTGACTG+AGCCTATC	4,566,180	0.98	97.68	2.32	1,379	100	92.13	35.61	NA	NA	NA

col <- as.character(as.numeric(grepl("Captive",ssWvCAV$Category)))
col <- gsub("0","lightgreen",col)
col <- gsub("1","lightblue",col)

mds<-plotMDS(x1,pch=19,cex=3,col=col,main="MDS plot: Wild vs Captive devils")
mtext("Wild=green, Captive=Blue")

ssWvCAV$Capstat <- grepl("Captive",ssWvCAV$Category)
ssWvCAV$MvF <- factor(ssWvCAV$Sex)
ssWvCAV$SubAdvAd <- as.numeric(factor(ssWvCAV$Age,levels=c("S","A"),ordered = TRUE))
ssWvCAV$Season <- factor(ssWvCAV$Season)
str(ssWvCAV)

## 'data.frame':    96 obs. of  34 variables:
##  $ Lane                   : int  1 1 1 1 1 1 1 1 1 1 ...
##  $ Project                : chr  "default" "default" "default" "default" ...
##  $ Client.ID              : chr  "S1_LauraGrace" "S1_Turmeric" "S2_Belinda" "S2_B99008" ...
##  $ Sample                 : int  6102461 6102465 6102467 6102468 6102469 6102470 6102475 6102476 6102477 6102484 ...
##  $ X                      : logi  NA NA NA NA NA NA ...
##  $ Category               : chr  "Healthy/Wild" "Healthy/Wild" "Healthy/Wild" "Captive" ...
##  $ Sex.Age.Season         : chr  "FAP" "FAU" "FAU" "FAU" ...
##  $ Sex                    : chr  "F" "F" "F" "F" ...
##  $ Age                    : chr  "A" "A" "A" "A" ...
##  $ Season                 : Factor w/ 4 levels "Autumn","Spring",..: 2 3 3 3 3 3 4 4 1 3 ...
##  $ Location               : chr  "Channel" "Channel" "Channel" "Monarto" ...
##  $ Agebyyear              : num  2.62 1.91 1.87 3.74 2.59 ...
##  $ Weight                 : chr  "7.2" "7.8" "6.6" "8.12" ...
##  $ X.1                    : logi  NA NA NA NA NA NA ...
##  $ Number.of.Tumours      : int  NA NA NA NA NA NA NA NA NA NA ...
##  $ Ulcerated              : logi  NA NA NA NA NA NA ...
##  $ UlcerationDepth        : chr  "" "" "" "" ...
##  $ Secondary.infection    : logi  NA NA NA NA NA NA ...
##  $ Total.tumour.volume    : num  NA NA NA NA NA NA NA NA NA NA ...
##  $ Barcode.sequence       : chr  "AGAGTAGC+AAGGCGTA" "GAAGGAAG+GACGTCAT" "GCAAGATC+AGTCGAAG" "GCTATCCT+AACAGGTG" ...
##  $ PF.Clusters            : chr  "4,511,702" "6,479,047" "3,296,547" "5,338,890" ...
##  $ X..of.the.lane         : num  1 1 1 1 3 1 1 1 3 1 ...
##  $ X..Perfect.barcode     : num  98.3 97.3 97.4 98.1 97.8 ...
##  $ X..One.mismatch.barcode: num  1.71 2.72 2.6 1.92 2.18 4.3 3.33 2.05 2.02 3.79 ...
##  $ Yield..Mbases.         : chr  "1,363" "1,957" "996" "1,612" ...
##  $ X..PF.Cluster          : int  100 100 100 100 100 100 100 100 100 100 ...
##  $ X.....Q30              : num  93.8 93.8 94.1 93.7 93.7 ...
##  $ Mean.Quality           : num  35.9 35.9 36 35.9 35.9 ...
##  $ X.2                    : logi  NA NA NA NA NA NA ...
##  $ X.3                    : logi  NA NA NA NA NA NA ...
##  $ X.4                    : logi  NA NA NA NA NA NA ...
##  $ Capstat                : logi  FALSE FALSE FALSE TRUE TRUE FALSE ...
##  $ MvF                    : Factor w/ 2 levels "F","M": 1 1 1 1 1 1 1 1 2 2 ...
##  $ SubAdvAd               : num  2 2 2 2 2 2 2 2 2 2 ...

dds <- DESeqDataSetFromMatrix(countData = x1 , colData = ssWvCAV, design = ~ Season + SubAdvAd + MvF + Capstat )

## converting counts to integer mode

##   the design formula contains one or more numeric variables with integer values,
##   specifying a model with increasing fold change for higher values.
##   did you mean for this to be a factor? if so, first convert
##   this variable to a factor using the factor() function

res <- DESeq(dds)

## estimating size factors

## estimating dispersions

## gene-wise dispersion estimates

## mean-dispersion relationship

## final dispersion estimates

## fitting model and testing

## -- replacing outliers and refitting for 8 genes
## -- DESeq argument 'minReplicatesForReplace' = 7 
## -- original counts are preserved in counts(dds)

## estimating dispersions

## fitting model and testing

z<- results(res)
vsd <- vst(dds, blind=FALSE,nsub=nrow(x1)-50)
zz<-cbind(as.data.frame(z),assay(vsd))
dge<-as.data.frame(zz[order(zz$pvalue),])
dgeWvCAV <- dge
sig <- subset(dge,padj<0.05)
sig1_up <- rownames(subset(sig,log2FoldChange>0))
sig1_dn <- rownames(subset(sig,log2FoldChange<0))
length(sig1_up)

## [1] 74

length(sig1_dn)

## [1] 68

maplot(dgeWvCAV,"Wild vs Captive devils")

make_volcano(dgeWvCAV,"Wild vs Captive devils")

sig[1:150,1:6] %>%
  kbl(caption="Comparison of Wild vs Captive devils") %>%
  kable_paper("hover", full_width = F)

Comparison of Wild vs Captive devils

	baseMean	log2FoldChange	lfcSE	stat	pvalue	padj
CCR7	1.767790e+04	1.2271697	0.1157199	10.604653	0.0000000	0.0000000
CD28	1.007240e+03	1.3905020	0.1484543	9.366536	0.0000000	0.0000000
CCR6	9.699889e+02	1.2372421	0.1326737	9.325453	0.0000000	0.0000000
TNFSF8	4.693606e+02	1.1511491	0.1307664	8.803095	0.0000000	0.0000000
Saha-UA	3.380555e+05	-1.2981149	0.1586440	-8.182566	0.0000000	0.0000000
PTPRC	3.958119e+04	1.0317417	0.1263583	8.165208	0.0000000	0.0000000
CXCR4	1.932450e+04	0.8056240	0.0989773	8.139481	0.0000000	0.0000000
CD1d	1.439792e+03	1.2576224	0.1552812	8.099000	0.0000000	0.0000000
IL10RA	1.142249e+04	1.1031254	0.1371707	8.041991	0.0000000	0.0000000
Saha-UK	2.183494e+05	-1.2089024	0.1503608	-8.040008	0.0000000	0.0000000
IL34	1.509099e+03	-1.8571371	0.2327797	-7.978089	0.0000000	0.0000000
CCL21	2.061270e+01	-6.4187162	0.8103442	-7.920975	0.0000000	0.0000000
Tap1	7.390331e+03	1.2820445	0.1622120	7.903511	0.0000000	0.0000000
TCF3	3.346812e+04	-1.1948086	0.1533903	-7.789335	0.0000000	0.0000000
EPX	3.946627e+03	-2.4814322	0.3195928	-7.764354	0.0000000	0.0000000
TLR3	5.283182e+02	0.9381514	0.1208608	7.762247	0.0000000	0.0000000
Saha-UC	5.052002e+05	-1.3170849	0.1698108	-7.756193	0.0000000	0.0000000
Saha-UD	1.028609e+06	-1.6972330	0.2205947	-7.693899	0.0000000	0.0000000
OAZ1	4.795188e+05	-1.4648367	0.1959791	-7.474454	0.0000000	0.0000000
LBR	9.041156e+03	0.9204106	0.1258553	7.313244	0.0000000	0.0000000
CCR4	1.319401e+03	0.9829831	0.1360250	7.226488	0.0000000	0.0000000
NFKB1	6.909295e+03	0.4549534	0.0630087	7.220485	0.0000000	0.0000000
TCRm	4.921642e+03	1.0222417	0.1416743	7.215437	0.0000000	0.0000000
ITGAX	6.031127e+03	-0.6612363	0.0918345	-7.200304	0.0000000	0.0000000
TGFb1	6.722329e+04	-0.9719383	0.1361172	-7.140451	0.0000000	0.0000000
IL10Rb	4.791712e+02	0.8730255	0.1242193	7.028099	0.0000000	0.0000000
STAT1	9.191622e+03	0.7508500	0.1090351	6.886316	0.0000000	0.0000000
IL7RA	6.847889e+03	0.9672449	0.1446519	6.686705	0.0000000	0.0000000
CCLD3	9.353998e+01	-2.3791038	0.3565814	-6.671979	0.0000000	0.0000000
IL2RA	2.234007e+02	1.1872530	0.1782697	6.659870	0.0000000	0.0000000
IL4R	6.524709e+03	0.8489715	0.1279479	6.635292	0.0000000	0.0000000
IgE	1.088380e+04	-1.7316618	0.2810735	-6.160885	0.0000000	0.0000000
GATA1	2.828681e+05	-2.0275587	0.3326270	-6.095592	0.0000000	0.0000000
TLR9	8.438999e+02	-0.9111577	0.1529552	-5.957022	0.0000000	0.0000000
Spi1	3.562727e+04	-0.8540119	0.1447617	-5.899433	0.0000000	0.0000000
IFIH1	1.955888e+03	0.7560749	0.1284833	5.884617	0.0000000	0.0000000
CD68	4.568176e+03	-0.7457534	0.1300779	-5.733129	0.0000000	0.0000001
TXLNA	4.446539e+03	-0.4643271	0.0825299	-5.626167	0.0000000	0.0000001
IRF8	1.753715e+03	0.8378874	0.1493430	5.610490	0.0000000	0.0000001
CIITA	1.709255e+03	0.7807442	0.1395429	5.595012	0.0000000	0.0000001
ICOS	5.716729e+02	0.8723781	0.1573448	5.544371	0.0000000	0.0000002
RPS29	2.719380e+04	-0.9794448	0.1772357	-5.526228	0.0000000	0.0000002
CCR8	1.850348e+02	1.0631347	0.1929769	5.509128	0.0000000	0.0000002
IL27Ra	1.216205e+03	0.8222411	0.1507574	5.454067	0.0000000	0.0000003
CCR2	4.309248e+03	-0.7255652	0.1339795	-5.415496	0.0000001	0.0000003
NLRP1	6.701689e+02	0.6887285	0.1273068	5.409988	0.0000001	0.0000003
GZMB	1.148661e+02	-1.2194297	0.2267256	-5.378440	0.0000001	0.0000004
IRF4	8.283538e+02	0.9532707	0.1780981	5.352505	0.0000001	0.0000005
RPLP0	1.679208e+05	-0.7973465	0.1491190	-5.347048	0.0000001	0.0000005
C3	4.582611e+04	-0.8216269	0.1542203	-5.327617	0.0000001	0.0000005
B2M	2.528038e+05	-0.9454463	0.1781885	-5.305876	0.0000001	0.0000006
Saha-MR1	2.499019e+02	0.6998754	0.1322361	5.292620	0.0000001	0.0000006
MYD88	5.384900e+03	0.7348046	0.1396863	5.260392	0.0000001	0.0000007
CXCL16	4.583658e+03	-0.9865119	0.1880591	-5.245755	0.0000002	0.0000007
IRF3	1.534376e+04	-0.7058176	0.1354579	-5.210606	0.0000002	0.0000009
IL5	2.703264e+02	2.9506055	0.5802679	5.084902	0.0000004	0.0000017
STAT3	7.656527e+03	0.6113772	0.1219203	5.014564	0.0000005	0.0000024
FLT3	1.402967e+03	0.8506604	0.1708691	4.978434	0.0000006	0.0000028
IL17RA	6.712231e+03	0.5871760	0.1219273	4.815786	0.0000015	0.0000063
PPIA	4.331074e+04	-0.6335900	0.1322751	-4.789941	0.0000017	0.0000071
STAT4	1.550386e+03	0.6420960	0.1345222	4.773159	0.0000018	0.0000076
CCR5	3.287396e+02	-0.6803694	0.1426450	-4.769670	0.0000018	0.0000076
TNFSF11	9.463621e+01	0.9907651	0.2098695	4.720862	0.0000023	0.0000095
IL5Ra	1.390130e+03	0.7261932	0.1554580	4.671315	0.0000030	0.0000119
CCLD14	2.527101e+02	-1.3662385	0.2928320	-4.665605	0.0000031	0.0000120
CCLD6	2.207176e+02	-0.9379357	0.2017742	-4.648441	0.0000033	0.0000129
TLR7	8.029070e+02	0.7555176	0.1636708	4.616079	0.0000039	0.0000148
CD59	2.008714e+04	-0.9962481	0.2172948	-4.584776	0.0000045	0.0000170
TCRd	2.667412e+02	0.9097317	0.1985964	4.580807	0.0000046	0.0000171
FGF2	6.673852e+01	0.9375102	0.2080684	4.505780	0.0000066	0.0000240
PRF1	1.789604e+03	0.8302502	0.1859961	4.463803	0.0000081	0.0000288
CCLD8	1.474595e+02	-0.8913450	0.2001414	-4.453577	0.0000084	0.0000298
CCLD5	4.730407e+02	-0.8420599	0.1902532	-4.425996	0.0000096	0.0000334
TCRg	2.489601e+03	-0.9398577	0.2126068	-4.420638	0.0000098	0.0000338
CCLD7	6.581604e+02	-0.8683252	0.1979701	-4.386142	0.0000115	0.0000391
CFI	5.246528e+02	0.4939781	0.1150217	4.294651	0.0000175	0.0000585
CD96	2.144166e+02	1.0580932	0.2474078	4.276718	0.0000190	0.0000626
TNFSF10	3.974518e+02	0.5064721	0.1186402	4.268974	0.0000196	0.0000639
KIT	3.189999e+03	0.6377188	0.1496893	4.260283	0.0000204	0.0000656
TNFSF13	5.170078e+02	-0.4802154	0.1147286	-4.185663	0.0000284	0.0000903
PRG3	9.291671e+03	0.8332135	0.2037356	4.089680	0.0000432	0.0001355
IL6	8.500763e+01	0.7378048	0.1842052	4.005342	0.0000619	0.0001918
CTLA4	2.653248e+02	0.5573867	0.1396224	3.992100	0.0000655	0.0002004
CD4	3.567471e+03	0.3524847	0.0885147	3.982217	0.0000683	0.0002065
CCL17	9.410251e+00	-3.2913368	0.8271205	-3.979271	0.0000691	0.0002066
IL13Ra1	3.102653e+03	0.6664027	0.1686731	3.950853	0.0000779	0.0002300
IL17D	2.486597e+02	0.8248296	0.2097502	3.932439	0.0000841	0.0002455
NOD2	1.330970e+03	0.7066856	0.1807951	3.908765	0.0000928	0.0002678
CCLD1	3.942714e+01	-1.6719777	0.4308829	-3.880353	0.0001043	0.0002977
VEGF-A	2.938225e+02	0.7097308	0.1843899	3.849076	0.0001186	0.0003346
GATA2	1.404719e+03	-0.5751660	0.1506359	-3.818252	0.0001344	0.0003751
AIM2	4.463174e+02	-0.4944053	0.1298133	-3.808589	0.0001398	0.0003812
IFNGR1	1.136194e+04	0.3649175	0.0958668	3.806506	0.0001409	0.0003812
STING1	9.494296e+02	0.4619155	0.1214383	3.803704	0.0001425	0.0003812
CCR1	4.428027e+02	-0.5949763	0.1564219	-3.803663	0.0001426	0.0003812
MMP2	6.184562e+01	-1.2668641	0.3357803	-3.772896	0.0001614	0.0004269
CD14	3.859311e+04	-0.6356577	0.1687375	-3.767139	0.0001651	0.0004324
IL4I0	1.167283e+02	-1.0237970	0.2745489	-3.729015	0.0001922	0.0004982
CD83	1.112830e+02	0.7355203	0.1990113	3.695871	0.0002191	0.0005622
Saha-UB	2.929331e+05	-0.6316950	0.1723459	-3.665275	0.0002471	0.0006276
IRF9	2.067724e+04	-0.2506027	0.0691738	-3.622801	0.0002914	0.0007329
FCGR2B	7.544089e+02	0.6658803	0.1849940	3.599470	0.0003189	0.0007940
FAS	1.207219e+03	0.4777228	0.1367249	3.494044	0.0004758	0.0011732
CCLD13	2.611072e+02	-0.7596304	0.2183393	-3.479129	0.0005030	0.0012176
EIF2AK2	5.172143e+02	0.4247775	0.1220985	3.478975	0.0005033	0.0012176
TLR8	5.257142e+03	0.4528657	0.1364788	3.318212	0.0009060	0.0021709
IL15	6.370138e+01	-1.0111239	0.3095030	-3.266928	0.0010872	0.0025809
TLR4	2.026238e+03	0.5583244	0.1716267	3.253133	0.0011414	0.0026844
IRF2	5.034485e+03	0.2034296	0.0635152	3.202849	0.0013608	0.0031709
TBX21	1.682176e+03	-0.4773975	0.1502198	-3.177994	0.0014830	0.0034243
CD209	1.970259e+01	-1.0888942	0.3447425	-3.158573	0.0015854	0.0036279
PDGFA	1.452147e+03	-0.7139984	0.2319908	-3.077701	0.0020860	0.0047308
CCLD2	1.408927e+01	-1.5992543	0.5287415	-3.024643	0.0024893	0.0055953
CXCR3	1.281664e+03	-0.4177267	0.1388065	-3.009418	0.0026175	0.0058133
IL27	3.762465e+02	0.3077496	0.1023192	3.007740	0.0026320	0.0058133
Saha-DAB3	8.420757e+04	-0.2600073	0.0871336	-2.984006	0.0028450	0.0062296
CCL24	1.131879e+03	-0.4965219	0.1669632	-2.973841	0.0029410	0.0063847
IL12Rb1	1.078035e+03	0.3652633	0.1240513	2.944454	0.0032353	0.0069640
CD40LG	9.253404e+02	0.2775242	0.0958157	2.896438	0.0037742	0.0080560
MRC1	1.246960e+03	-0.4707819	0.1653583	-2.847041	0.0044128	0.0093404
IFNAR1	2.921068e+03	0.2710338	0.0969812	2.794704	0.0051947	0.0109046
Saha-DAB2	1.377242e+04	-0.2773343	0.1009608	-2.746949	0.0060152	0.0125236
CD86	3.089243e+02	-0.3597305	0.1313234	-2.739273	0.0061575	0.0127155
TNFSF15	1.498152e+01	1.2243209	0.4509685	2.714870	0.0066302	0.0135812
XCLA	8.400273e+02	-0.3772685	0.1404018	-2.687063	0.0072083	0.0146473
Tap2	1.944570e+04	-0.2328117	0.0873179	-2.666254	0.0076702	0.0154621
RPL4	2.089100e+05	-0.2639778	0.1012556	-2.607044	0.0091328	0.0182655
FCER1A	2.737214e+02	-0.4957212	0.1913086	-2.591213	0.0095638	0.0189782
IL21R	4.836290e+03	0.3202950	0.1259769	2.542490	0.0110066	0.0216718
EEF1A1	3.186484e+01	0.8245803	0.3273708	2.518796	0.0117757	0.0229685
NLCR5	8.129648e+03	0.2483898	0.0986966	2.516700	0.0118460	0.0229685
CXCR5	2.227051e+03	0.3418899	0.1382471	2.473035	0.0133971	0.0257792
MS4A1	1.214175e+04	-0.3519964	0.1460125	-2.410728	0.0159207	0.0304049
IL16	1.891656e+04	0.1320063	0.0563511	2.342567	0.0191516	0.0363023
CCLD9	4.575973e+01	-0.6005962	0.2574661	-2.332719	0.0196629	0.0369954
FCMR	5.396858e+04	-0.3973649	0.1746454	-2.275267	0.0228899	0.0427503
IFNL2	2.660477e+01	-0.8378638	0.3702900	-2.262723	0.0236528	0.0438526
IDO1	2.441286e+03	0.9196361	0.4075943	2.256254	0.0240547	0.0442747
Saha-DAA	8.958693e+04	0.2047506	0.0916460	2.234147	0.0254734	0.0465485
TNFSF13B	5.193844e+01	-0.5561523	0.2498534	-2.225915	0.0260199	0.0472075
CSF1	1.065547e+03	0.4711923	0.2125562	2.216789	0.0266375	0.0479852
TNFSF9	2.951905e+02	-0.3953574	0.1798061	-2.198799	0.0278922	0.0498918
NA	NA	NA	NA	NA	NA	NA
NA.1	NA	NA	NA	NA	NA	NA
NA.2	NA	NA	NA	NA	NA	NA
NA.3	NA	NA	NA	NA	NA	NA
NA.4	NA	NA	NA	NA	NA	NA
NA.5	NA	NA	NA	NA	NA	NA
NA.6	NA	NA	NA	NA	NA	NA
NA.7	NA	NA	NA	NA	NA	NA

write.table(dge,file="dgeWvCAV.tsv",sep="\t")

mx <- sig[,7:ncol(sig)]
mx <- head(mx,30)
colfunc <- colorRampPalette(c("blue", "white", "red"))
heatmap.2(as.matrix(mx),trace="none",scale="row",
          col=colfunc(25),ColSideColors=col,mar=c(5,12)) 

ssCap <- filter(ss, Category %in% c('Captive'))

ssCLAV <- ssCap[which(grepl("Healesville",ssCap$Location) + grepl("Monarto",ssCap$Location)>0),]
ssCLAV

##    Lane Project      Client.ID  Sample  X Category Sex.Age.Season Sex Age
## 1     1 default      S2_B99008 6102468 NA  Captive            FAU   F   A
## 2     1 default      S2_Bonney 6102469 NA  Captive            FAU   F   A
## 3     1 default      S3_Jessie 6102476 NA  Captive            FAW   F   A
## 4     1 default    S3_McCauley 6102477 NA  Captive            MAA   M   A
## 5     1 default      S5_Fenton 6102484 NA  Captive            MAU   M   A
## 6     1 default S5_Bryan Adams 6102485 NA  Captive            MAW   M   A
## 7     1 default       S7_Fredo 6102496 NA  Captive            MAA   M   A
## 8     1 default      S7_Piglet 6102497 NA  Captive            FSU   F   S
## 9     1 default    S10_McMahon 6102514 NA  Captive            MSP   M   S
## 10    1 default      S10_Logan 6102515 NA  Captive            MAU   M   A
## 11    1 default       S11_Xuki 6102520 NA  Captive            MAA   M   A
## 12    1 default   S11_Kalamata 6102521 NA  Captive            MSW   M   S
## 13    1 default      S12_Wally 6102528 NA  Captive            MAA   M   A
## 14    1 default        S12_Naz 6102529 NA  Captive            MAW   M   A
## 15    1 default     S13_Humbug 6102530 NA  Captive            MAU   M   A
## 16    1 default     S13_Xerxes 6102535 NA  Captive            MAA   M   A
## 17    1 default     S14_Catani 6102538 NA  Captive            FAU   F   A
## 18    2 default     S15_Amelia 6102541 NA  Captive            FAU   F   A
## 19    2 default     S15_London 6102542 NA  Captive            FAU   F   A
## 20    2 default      S15_Elyse 6102543 NA  Captive            FAP   F   A
## 21    2 default     S16_Marley 6102547 NA  Captive            FAU   F   A
## 22    2 default      S16_Milla 6102548 NA  Captive            FAU   F   A
## 23    2 default S16_Maleficent 6102549 NA  Captive            FAU   F   A
## 24    2 default     S17_Anirak 6102553 NA  Captive            FSW   F   S
## 25    2 default       S17_Poly 6102554 NA  Captive            FSU   F   S
## 26    2 default     S17_Methyl 6102556 NA  Captive            FSU   F   S
## 27    2 default       S18_Zara 6102558 NA  Captive            FSW   F   S
## 28    2 default       S18_Maze 6102559 NA  Captive            FSW   F   S
## 29    2 default    S18_Delores 6102560 NA  Captive            FSW   F   S
## 30    2 default    S19_Morocco 6102564 NA  Captive            MAU   M   A
## 31    2 default   S19_Pecorino 6102565 NA  Captive            FSW   F   S
## 32    2 default      S19_Hooch 6102566 NA  Captive            MSA   M   S
## 33    2 default      S20_Muasy 6102569 NA  Captive            MSU   M   S
## 34    2 default       S20_Fuse 6102570 NA  Captive            MSU   M   S
## 35    2 default     S20_Joseph 6102571 NA  Captive            MSU   M   S
## 36    2 default      S21_Moama 6102576 NA  Captive            MAW   M   A
## 37    2 default     S21_Nauset 6102577 NA  Captive            MSW   M   S
## 38    2 default     S21_Graham 6102578 NA  Captive            MSW   M   S
## 39    2 default      S22_Yanni 6102581 NA  Captive            MSW   M   S
## 40    2 default    S22_Gherkin 6102582 NA  Captive            MSW   M   S
## 41    2 default    S22_Stilton 6102584 NA  Captive            MSW   M   S
## 42    2 default     S23_Osiris 6102588 NA  Captive            MSW   M   S
## 43    2 default     S23_X-boxx 6102589 NA  Captive            FSW   F   S
## 44    2 default     S23_Quince 6102590 NA  Captive            MSW   M   S
## 45    2 default     S24_Malfoy 6102595 NA  Captive            MAW   M   A
## 46    2 default   S24_Licorice 6102596 NA  Captive            FSA   F   S
## 47    2 default       S25_Tofu 6102599 NA  Captive            FAU   F   A
## 48    2 default    S25_Colette 6102600 NA  Captive            FAP   F   A
## 49    2 default S25_CannonBall 6102601 NA  Captive            MSW   M   S
## 50    2 default  S25_Cornelius 6102602 NA  Captive            MAU   M   A
## 51    2 default    S25_Mardook 6102603 NA  Captive            MAW   M   A
## 52    2 default    S25_Negroni 6102604 NA  Captive            FSP   F   S
##    Season    Location Agebyyear Weight X.1 Number.of.Tumours Ulcerated
## 1  Summer     Monarto 3.7369863   8.12  NA                NA        NA
## 2  Summer Healesville 2.5863014    5.4  NA                NA        NA
## 3  Winter     Monarto 2.1945205  12.95  NA                NA        NA
## 4  Autumn Healesville 3.0328767   8.52  NA                NA        NA
## 5  Summer     Monarto 3.0000000    N/A  NA                NA        NA
## 6  Winter Healesville 4.1616438    8.6  NA                NA        NA
## 7  Autumn Healesville 2.9945205    8.8  NA                NA        NA
## 8  Summer Healesville 1.5890411   4.76  NA                NA        NA
## 9  Spring Healesville 1.6493151   8.06  NA                NA        NA
## 10 Summer Healesville 5.6000000   9.02  NA                NA        NA
## 11 Autumn     Monarto 2.0767123   7.95  NA                NA        NA
## 12 Winter Healesville 1.2849315   5.58  NA                NA        NA
## 13 Autumn     Monarto 2.1616438    N/A  NA                NA        NA
## 14 Winter     Monarto 4.4547945    N/A  NA                NA        NA
## 15 Summer     Monarto 2.5397260   7.94  NA                NA        NA
## 16 Autumn     Monarto 2.0767123    N/A  NA                NA        NA
## 17 Summer Healesville 2.8630137   6.46  NA                NA        NA
## 18 Summer Healesville 4.7260274   5.85  NA                NA        NA
## 19 Summer Healesville 3.7863014    4.8  NA                NA        NA
## 20 Spring Healesville 4.5232877   7.88  NA                NA        NA
## 21 Summer     Monarto 4.7726027   7.51  NA                NA        NA
## 22 Summer Healesville 2.7671233   5.82  NA                NA        NA
## 23 Summer     Monarto 4.5589041   8.59  NA                NA        NA
## 24 Winter Healesville 1.3424658   5.42  NA                NA        NA
## 25 Summer     Monarto 1.7917808   6.67  NA                NA        NA
## 26 Summer Healesville 0.7534247   2.72  NA                NA        NA
## 27 Winter     Monarto 1.5150685    4.6  NA                NA        NA
## 28 Winter Healesville 1.4000000   5.82  NA                NA        NA
## 29 Winter Healesville 1.2986301    5.1  NA                NA        NA
## 30 Summer     Monarto 4.5917808   9.82  NA                NA        NA
## 31 Winter Healesville 1.1123288   6.02  NA                NA        NA
## 32 Autumn Healesville 1.0164384    6.1  NA                NA        NA
## 33 Summer Healesville 1.4219178   6.62  NA                NA        NA
## 34 Summer Healesville 1.7315068   6.26  NA                NA        NA
## 35 Summer Healesville 1.7479452   7.42  NA                NA        NA
## 36 Winter     Monarto 1.6986301   8.75  NA                NA        NA
## 37 Winter Healesville 1.3041096  4.036  NA                NA        NA
## 38 Winter Healesville 1.3616438    6.4  NA                NA        NA
## 39 Winter     Monarto 1.5178082   8.15  NA                NA        NA
## 40 Winter Healesville 1.2630137   6.32  NA                NA        NA
## 41 Winter Healesville 1.2986301   5.76  NA                NA        NA
## 42 Winter Healesville 1.1753425   5.34  NA                NA        NA
## 43 Winter     Monarto 1.4657534   4.34  NA                NA        NA
## 44 Winter Healesville 1.3835616    N/A  NA                NA        NA
## 45 Winter Healesville 5.1041096   8.02  NA                NA        NA
## 46 Autumn Healesville 1.8246575   5.94  NA                NA        NA
## 47 Summer Healesville 2.7452055    N/A  NA                NA        NA
## 48 Spring Healesville 6.6273973   7.52  NA                NA        NA
## 49 Winter Healesville 1.1479452   6.54  NA                NA        NA
## 50 Summer Healesville 3.6904110   7.74  NA                NA        NA
## 51 Winter Healesville 4.3013699   9.16  NA                NA        NA
## 52 Spring Healesville 1.6767123   5.86  NA                NA        NA
##    UlcerationDepth Secondary.infection Total.tumour.volume  Barcode.sequence
## 1                                   NA                  NA GCTATCCT+AACAGGTG
## 2                                   NA                  NA ACGTTCAG+AGTTGTGC
## 3                                   NA                  NA CCGTAAGA+AGAGTCCA
## 4                                   NA                  NA CAACACCT+AGCTACCA
## 5                                   NA                  NA GATTACCG+TTGGACTG
## 6                                   NA                  NA CACTAGCT+TCGGATTC
## 7                                   NA                  NA AGTGGATC+TCGATGAC
## 8                                   NA                  NA ATCACACG+GTAGCGTA
## 9                                   NA                  NA AACGTCTG+AATGACGC
## 10                                  NA                  NA GTGCCATA+CTCCTAGT
## 11                                  NA                  NA ATCGCCAT+TGCGTAAC
## 12                                  NA                  NA GAACATCG+TCGACAAG
## 13                                  NA                  NA ACGACTTG+AACAGCGA
## 14                                  NA                  NA CCTTGTAG+CCTTGGAA
## 15                                  NA                  NA GAGCAGTA+TGAGCTGT
## 16                                  NA                  NA GATAGGCT+CAGTCACA
## 17                                  NA                  NA TGAAGACG+ACATGCCA
## 18                                  NA                  NA AACGTCTG+AATGACGC
## 19                                  NA                  NA GTCATCGA+GATCCACT
## 20                                  NA                  NA AGCGTGTT+CTTCGCAA
## 21                                  NA                  NA GCAAGATC+AGTCGAAG
## 22                                  NA                  NA TTGCGAAG+AACACGCT
## 23                                  NA                  NA TGCTTCCA+CGATCGAT
## 24                                  NA                  NA TGAGCTAG+GAACGGTT
## 25                                  NA                  NA CTTAGGAC+ACTCCTAC
## 26                                  NA                  NA TGGTACAG+TCGAGAGT
## 27                                  NA                  NA ACCACGAT+GTCTGCAA
## 28                                  NA                  NA ACTGAGGT+GTCAACAG
## 29                                  NA                  NA ACGACTTG+AACAGCGA
## 30                                  NA                  NA TCAGGCTT+ATCATGCG
## 31                                  NA                  NA ATGCCTGT+AGATTGCG
## 32                                  NA                  NA CCGTAAGA+AGAGTCCA
## 33                                  NA                  NA TTCCAAGG+CTACAAGG
## 34                                  NA                  NA GTAGGAGT+GTCCTAAG
## 35                                  NA                  NA CTTCGACT+GATCTTGC
## 36                                  NA                  NA TTGGACGT+GAAGTGCT
## 37                                  NA                  NA TATCGGTC+TACCGGAT
## 38                                  NA                  NA TACGCCTT+GCTACTCT
## 39                                  NA                  NA CGTGTGTA+CCAACGAA
## 40                                  NA                  NA CCTTGTAG+CCTTGGAA
## 41                                  NA                  NA CGCAATCT+ACAGGCAT
## 42                                  NA                  NA AGTGGATC+TCGATGAC
## 43                                  NA                  NA CTGTTGAC+ACCTCAGT
## 44                                  NA                  NA AACCGTTC+CTAGCTCA
## 45                                  NA                  NA TAACCGGT+ATCGTCTC
## 46                                  NA                  NA TTGCAGAC+ATCGTGGT
## 47                                  NA                  NA GAGACGAT+ACCGGTTA
## 48                                  NA                  NA CTTGTCGA+CGATGTTC
## 49                                  NA                  NA TACGCTAC+CGTGTGAT
## 50                                  NA                  NA GAATCCGA+AGCTAGTG
## 51                                  NA                  NA GCACAACT+CTGAACGT
## 52                                  NA                  NA TGTGACTG+AGCCTATC
##    PF.Clusters X..of.the.lane X..Perfect.barcode X..One.mismatch.barcode
## 1    5,338,890           1.00              98.08                    1.92
## 2   14,968,459           3.00              97.82                    2.18
## 3    4,340,498           1.00              97.95                    2.05
## 4   16,944,152           3.00              97.98                    2.02
## 5    5,125,566           1.00              96.21                    3.79
## 6    5,317,767           1.00              97.33                    2.67
## 7    2,690,632           1.00              97.77                    2.23
## 8   12,134,750           2.00              96.94                    3.06
## 9    3,809,607           1.00              96.73                    3.27
## 10   9,606,720           2.00              97.52                    2.48
## 11   6,018,884           1.00              97.88                    2.12
## 12   6,861,544           1.00              97.76                    2.24
## 13   3,000,501           1.00              97.96                    2.04
## 14   5,927,942           1.00              97.40                    2.60
## 15   1,702,608           0.00              98.07                    1.93
## 16   7,628,257           2.00              98.34                    1.66
## 17  13,381,058           3.00              96.98                    3.02
## 18   8,041,818           1.73              96.32                    3.68
## 19   5,758,111           1.24              96.72                    3.28
## 20   5,774,637           1.24              96.76                    3.24
## 21     348,621           0.08              97.24                    2.76
## 22   6,376,997           1.37              97.60                    2.40
## 23   9,181,668           1.98              98.08                    1.92
## 24   3,740,878           0.81              97.15                    2.85
## 25  11,133,566           2.40              97.85                    2.15
## 26   7,887,375           1.70              97.57                    2.43
## 27   7,308,907           1.58              97.70                    2.30
## 28   2,984,930           0.64              97.59                    2.41
## 29          98           0.00              94.90                    5.10
## 30   4,879,629           1.05              97.38                    2.62
## 31   4,857,395           1.05              97.58                    2.42
## 32   1,074,345           0.23              97.56                    2.44
## 33  10,669,237           2.30              97.64                    2.36
## 34  10,603,179           2.29              97.34                    2.66
## 35   5,642,025           1.22              96.37                    3.63
## 36  10,821,895           2.33              96.02                    3.98
## 37   8,693,920           1.87              98.13                    1.87
## 38   3,422,010           0.74              97.72                    2.28
## 39   4,811,453           1.04              97.67                    2.33
## 40   5,913,586           1.27              97.05                    2.95
## 41   8,306,537           1.79              96.94                    3.06
## 42   3,528,471           0.76              97.24                    2.76
## 43          23           0.00             100.00                     NaN
## 44   7,247,793           1.56              97.63                    2.37
## 45   6,908,416           1.49              97.09                    2.91
## 46   4,475,461           0.96              95.05                    4.95
## 47  12,024,421           2.59              95.37                    4.63
## 48  10,751,797           2.32              97.49                    2.51
## 49  13,015,081           2.81              97.36                    2.64
## 50   3,336,983           0.72              97.10                    2.90
## 51   6,826,313           1.47              97.66                    2.34
## 52   4,566,180           0.98              97.68                    2.32
##    Yield..Mbases. X..PF.Cluster X.....Q30 Mean.Quality X.2 X.3 X.4
## 1           1,612           100     93.69        35.91  NA  NA  NA
## 2           4,520           100     93.68        35.91  NA  NA  NA
## 3           1,311           100     93.54        35.88  NA  NA  NA
## 4           5,117           100     93.05        35.78  NA  NA  NA
## 5           1,548           100     93.89        35.94  NA  NA  NA
## 6           1,606           100     93.24        35.83  NA  NA  NA
## 7             813           100     93.77        35.93  NA  NA  NA
## 8           3,665           100     93.34        35.84  NA  NA  NA
## 9           1,151           100     93.07        35.79  NA  NA  NA
## 10          2,901           100     93.25        35.83  NA  NA  NA
## 11          1,818           100     93.58        35.89  NA  NA  NA
## 12          2,072           100     93.49        35.87  NA  NA  NA
## 13            906           100     93.70        35.91  NA  NA  NA
## 14          1,790           100     93.86        35.94  NA  NA  NA
## 15            514           100     93.58        35.89  NA  NA  NA
## 16          2,304           100     93.62        35.89  NA  NA  NA
## 17          4,041           100     93.36        35.84  NA  NA  NA
## 18          2,429           100     92.18        35.63  NA  NA  NA
## 19          1,739           100     91.87        35.55  NA  NA  NA
## 20          1,744           100     91.56        35.50  NA  NA  NA
## 21            105           100     92.77        35.73  NA  NA  NA
## 22          1,926           100     92.44        35.67  NA  NA  NA
## 23          2,773           100     93.04        35.79  NA  NA  NA
## 24          1,130           100     93.03        35.79  NA  NA  NA
## 25          3,362           100     92.47        35.68  NA  NA  NA
## 26          2,382           100     93.15        35.81  NA  NA  NA
## 27          2,207           100     92.84        35.75  NA  NA  NA
## 28            901           100     92.49        35.69  NA  NA  NA
## 29              0           100     89.11        34.91  NA  NA  NA
## 30          1,474           100     92.63        35.71  NA  NA  NA
## 31          1,467           100     92.64        35.72  NA  NA  NA
## 32            324           100     92.66        35.72  NA  NA  NA
## 33          3,222           100     92.79        35.74  NA  NA  NA
## 34          3,202           100     92.01        35.58  NA  NA  NA
## 35          1,704           100     91.87        35.56  NA  NA  NA
## 36          3,268           100     92.72        35.73  NA  NA  NA
## 37          2,626           100     92.32        35.65  NA  NA  NA
## 38          1,033           100     92.94        35.76  NA  NA  NA
## 39          1,453           100     92.30        35.65  NA  NA  NA
## 40          1,786           100     92.92        35.77  NA  NA  NA
## 41          2,509           100     91.86        35.55  NA  NA  NA
## 42          1,066           100     93.08        35.79  NA  NA  NA
## 43              0           100     95.85        36.37  NA  NA  NA
## 44          2,189           100     92.28        35.63  NA  NA  NA
## 45          2,086           100     91.90        35.56  NA  NA  NA
## 46          1,352           100     91.42        35.46  NA  NA  NA
## 47          3,631           100     93.06        35.79  NA  NA  NA
## 48          3,247           100     92.77        35.74  NA  NA  NA
## 49          3,931           100     92.61        35.70  NA  NA  NA
## 50          1,008           100     93.09        35.79  NA  NA  NA
## 51          2,062           100     92.67        35.71  NA  NA  NA
## 52          1,379           100     92.13        35.61  NA  NA  NA

ssCLAV <- ssCLAV[ssCLAV$Sample %in% colnames(x),]

dim(ssCLAV)

## [1] 48 31

x1 <- x[,which(colnames(x) %in% ssCLAV$Sample)]
dim(x1)

## [1] 310  48

x1 <- x1[which(rowMeans(x1)>10),]
dim(x1)

## [1] 261  48

ssCLAV %>%  kbl(caption="Sample sheet")%>% kable_styling("hover",full_width=FALSE) 

Sample sheet

	Lane	Project	Client.ID	Sample	X	Category	Sex.Age.Season	Sex	Age	Season	Location	Agebyyear	Weight	X.1	Number.of.Tumours	Ulcerated	UlcerationDepth	Secondary.infection	Total.tumour.volume	Barcode.sequence	PF.Clusters	X..of.the.lane	X..Perfect.barcode	X..One.mismatch.barcode	Yield..Mbases.	X..PF.Cluster	X…..Q30	Mean.Quality	X.2	X.3	X.4
1	1	default	S2_B99008	6102468	NA	Captive	FAU	F	A	Summer	Monarto	3.7369863	8.12	NA	NA	NA		NA	NA	GCTATCCT+AACAGGTG	5,338,890	1.00	98.08	1.92	1,612	100	93.69	35.91	NA	NA	NA
2	1	default	S2_Bonney	6102469	NA	Captive	FAU	F	A	Summer	Healesville	2.5863014	5.4	NA	NA	NA		NA	NA	ACGTTCAG+AGTTGTGC	14,968,459	3.00	97.82	2.18	4,520	100	93.68	35.91	NA	NA	NA
3	1	default	S3_Jessie	6102476	NA	Captive	FAW	F	A	Winter	Monarto	2.1945205	12.95	NA	NA	NA		NA	NA	CCGTAAGA+AGAGTCCA	4,340,498	1.00	97.95	2.05	1,311	100	93.54	35.88	NA	NA	NA
4	1	default	S3_McCauley	6102477	NA	Captive	MAA	M	A	Autumn	Healesville	3.0328767	8.52	NA	NA	NA		NA	NA	CAACACCT+AGCTACCA	16,944,152	3.00	97.98	2.02	5,117	100	93.05	35.78	NA	NA	NA
5	1	default	S5_Fenton	6102484	NA	Captive	MAU	M	A	Summer	Monarto	3.0000000	N/A	NA	NA	NA		NA	NA	GATTACCG+TTGGACTG	5,125,566	1.00	96.21	3.79	1,548	100	93.89	35.94	NA	NA	NA
6	1	default	S5_Bryan Adams	6102485	NA	Captive	MAW	M	A	Winter	Healesville	4.1616438	8.6	NA	NA	NA		NA	NA	CACTAGCT+TCGGATTC	5,317,767	1.00	97.33	2.67	1,606	100	93.24	35.83	NA	NA	NA
7	1	default	S7_Fredo	6102496	NA	Captive	MAA	M	A	Autumn	Healesville	2.9945205	8.8	NA	NA	NA		NA	NA	AGTGGATC+TCGATGAC	2,690,632	1.00	97.77	2.23	813	100	93.77	35.93	NA	NA	NA
8	1	default	S7_Piglet	6102497	NA	Captive	FSU	F	S	Summer	Healesville	1.5890411	4.76	NA	NA	NA		NA	NA	ATCACACG+GTAGCGTA	12,134,750	2.00	96.94	3.06	3,665	100	93.34	35.84	NA	NA	NA
9	1	default	S10_McMahon	6102514	NA	Captive	MSP	M	S	Spring	Healesville	1.6493151	8.06	NA	NA	NA		NA	NA	AACGTCTG+AATGACGC	3,809,607	1.00	96.73	3.27	1,151	100	93.07	35.79	NA	NA	NA
10	1	default	S10_Logan	6102515	NA	Captive	MAU	M	A	Summer	Healesville	5.6000000	9.02	NA	NA	NA		NA	NA	GTGCCATA+CTCCTAGT	9,606,720	2.00	97.52	2.48	2,901	100	93.25	35.83	NA	NA	NA
11	1	default	S11_Xuki	6102520	NA	Captive	MAA	M	A	Autumn	Monarto	2.0767123	7.95	NA	NA	NA		NA	NA	ATCGCCAT+TGCGTAAC	6,018,884	1.00	97.88	2.12	1,818	100	93.58	35.89	NA	NA	NA
12	1	default	S11_Kalamata	6102521	NA	Captive	MSW	M	S	Winter	Healesville	1.2849315	5.58	NA	NA	NA		NA	NA	GAACATCG+TCGACAAG	6,861,544	1.00	97.76	2.24	2,072	100	93.49	35.87	NA	NA	NA
13	1	default	S12_Wally	6102528	NA	Captive	MAA	M	A	Autumn	Monarto	2.1616438	N/A	NA	NA	NA		NA	NA	ACGACTTG+AACAGCGA	3,000,501	1.00	97.96	2.04	906	100	93.70	35.91	NA	NA	NA
14	1	default	S12_Naz	6102529	NA	Captive	MAW	M	A	Winter	Monarto	4.4547945	N/A	NA	NA	NA		NA	NA	CCTTGTAG+CCTTGGAA	5,927,942	1.00	97.40	2.60	1,790	100	93.86	35.94	NA	NA	NA
15	1	default	S13_Humbug	6102530	NA	Captive	MAU	M	A	Summer	Monarto	2.5397260	7.94	NA	NA	NA		NA	NA	GAGCAGTA+TGAGCTGT	1,702,608	0.00	98.07	1.93	514	100	93.58	35.89	NA	NA	NA
16	1	default	S13_Xerxes	6102535	NA	Captive	MAA	M	A	Autumn	Monarto	2.0767123	N/A	NA	NA	NA		NA	NA	GATAGGCT+CAGTCACA	7,628,257	2.00	98.34	1.66	2,304	100	93.62	35.89	NA	NA	NA
17	1	default	S14_Catani	6102538	NA	Captive	FAU	F	A	Summer	Healesville	2.8630137	6.46	NA	NA	NA		NA	NA	TGAAGACG+ACATGCCA	13,381,058	3.00	96.98	3.02	4,041	100	93.36	35.84	NA	NA	NA
18	2	default	S15_Amelia	6102541	NA	Captive	FAU	F	A	Summer	Healesville	4.7260274	5.85	NA	NA	NA		NA	NA	AACGTCTG+AATGACGC	8,041,818	1.73	96.32	3.68	2,429	100	92.18	35.63	NA	NA	NA
19	2	default	S15_London	6102542	NA	Captive	FAU	F	A	Summer	Healesville	3.7863014	4.8	NA	NA	NA		NA	NA	GTCATCGA+GATCCACT	5,758,111	1.24	96.72	3.28	1,739	100	91.87	35.55	NA	NA	NA
20	2	default	S15_Elyse	6102543	NA	Captive	FAP	F	A	Spring	Healesville	4.5232877	7.88	NA	NA	NA		NA	NA	AGCGTGTT+CTTCGCAA	5,774,637	1.24	96.76	3.24	1,744	100	91.56	35.50	NA	NA	NA
22	2	default	S16_Milla	6102548	NA	Captive	FAU	F	A	Summer	Healesville	2.7671233	5.82	NA	NA	NA		NA	NA	TTGCGAAG+AACACGCT	6,376,997	1.37	97.60	2.40	1,926	100	92.44	35.67	NA	NA	NA
23	2	default	S16_Maleficent	6102549	NA	Captive	FAU	F	A	Summer	Monarto	4.5589041	8.59	NA	NA	NA		NA	NA	TGCTTCCA+CGATCGAT	9,181,668	1.98	98.08	1.92	2,773	100	93.04	35.79	NA	NA	NA
24	2	default	S17_Anirak	6102553	NA	Captive	FSW	F	S	Winter	Healesville	1.3424658	5.42	NA	NA	NA		NA	NA	TGAGCTAG+GAACGGTT	3,740,878	0.81	97.15	2.85	1,130	100	93.03	35.79	NA	NA	NA
25	2	default	S17_Poly	6102554	NA	Captive	FSU	F	S	Summer	Monarto	1.7917808	6.67	NA	NA	NA		NA	NA	CTTAGGAC+ACTCCTAC	11,133,566	2.40	97.85	2.15	3,362	100	92.47	35.68	NA	NA	NA
26	2	default	S17_Methyl	6102556	NA	Captive	FSU	F	S	Summer	Healesville	0.7534247	2.72	NA	NA	NA		NA	NA	TGGTACAG+TCGAGAGT	7,887,375	1.70	97.57	2.43	2,382	100	93.15	35.81	NA	NA	NA
27	2	default	S18_Zara	6102558	NA	Captive	FSW	F	S	Winter	Monarto	1.5150685	4.6	NA	NA	NA		NA	NA	ACCACGAT+GTCTGCAA	7,308,907	1.58	97.70	2.30	2,207	100	92.84	35.75	NA	NA	NA
28	2	default	S18_Maze	6102559	NA	Captive	FSW	F	S	Winter	Healesville	1.4000000	5.82	NA	NA	NA		NA	NA	ACTGAGGT+GTCAACAG	2,984,930	0.64	97.59	2.41	901	100	92.49	35.69	NA	NA	NA
30	2	default	S19_Morocco	6102564	NA	Captive	MAU	M	A	Summer	Monarto	4.5917808	9.82	NA	NA	NA		NA	NA	TCAGGCTT+ATCATGCG	4,879,629	1.05	97.38	2.62	1,474	100	92.63	35.71	NA	NA	NA
31	2	default	S19_Pecorino	6102565	NA	Captive	FSW	F	S	Winter	Healesville	1.1123288	6.02	NA	NA	NA		NA	NA	ATGCCTGT+AGATTGCG	4,857,395	1.05	97.58	2.42	1,467	100	92.64	35.72	NA	NA	NA
33	2	default	S20_Muasy	6102569	NA	Captive	MSU	M	S	Summer	Healesville	1.4219178	6.62	NA	NA	NA		NA	NA	TTCCAAGG+CTACAAGG	10,669,237	2.30	97.64	2.36	3,222	100	92.79	35.74	NA	NA	NA
34	2	default	S20_Fuse	6102570	NA	Captive	MSU	M	S	Summer	Healesville	1.7315068	6.26	NA	NA	NA		NA	NA	GTAGGAGT+GTCCTAAG	10,603,179	2.29	97.34	2.66	3,202	100	92.01	35.58	NA	NA	NA
35	2	default	S20_Joseph	6102571	NA	Captive	MSU	M	S	Summer	Healesville	1.7479452	7.42	NA	NA	NA		NA	NA	CTTCGACT+GATCTTGC	5,642,025	1.22	96.37	3.63	1,704	100	91.87	35.56	NA	NA	NA
36	2	default	S21_Moama	6102576	NA	Captive	MAW	M	A	Winter	Monarto	1.6986301	8.75	NA	NA	NA		NA	NA	TTGGACGT+GAAGTGCT	10,821,895	2.33	96.02	3.98	3,268	100	92.72	35.73	NA	NA	NA
37	2	default	S21_Nauset	6102577	NA	Captive	MSW	M	S	Winter	Healesville	1.3041096	4.036	NA	NA	NA		NA	NA	TATCGGTC+TACCGGAT	8,693,920	1.87	98.13	1.87	2,626	100	92.32	35.65	NA	NA	NA
38	2	default	S21_Graham	6102578	NA	Captive	MSW	M	S	Winter	Healesville	1.3616438	6.4	NA	NA	NA		NA	NA	TACGCCTT+GCTACTCT	3,422,010	0.74	97.72	2.28	1,033	100	92.94	35.76	NA	NA	NA
39	2	default	S22_Yanni	6102581	NA	Captive	MSW	M	S	Winter	Monarto	1.5178082	8.15	NA	NA	NA		NA	NA	CGTGTGTA+CCAACGAA	4,811,453	1.04	97.67	2.33	1,453	100	92.30	35.65	NA	NA	NA
40	2	default	S22_Gherkin	6102582	NA	Captive	MSW	M	S	Winter	Healesville	1.2630137	6.32	NA	NA	NA		NA	NA	CCTTGTAG+CCTTGGAA	5,913,586	1.27	97.05	2.95	1,786	100	92.92	35.77	NA	NA	NA
41	2	default	S22_Stilton	6102584	NA	Captive	MSW	M	S	Winter	Healesville	1.2986301	5.76	NA	NA	NA		NA	NA	CGCAATCT+ACAGGCAT	8,306,537	1.79	96.94	3.06	2,509	100	91.86	35.55	NA	NA	NA
42	2	default	S23_Osiris	6102588	NA	Captive	MSW	M	S	Winter	Healesville	1.1753425	5.34	NA	NA	NA		NA	NA	AGTGGATC+TCGATGAC	3,528,471	0.76	97.24	2.76	1,066	100	93.08	35.79	NA	NA	NA
44	2	default	S23_Quince	6102590	NA	Captive	MSW	M	S	Winter	Healesville	1.3835616	N/A	NA	NA	NA		NA	NA	AACCGTTC+CTAGCTCA	7,247,793	1.56	97.63	2.37	2,189	100	92.28	35.63	NA	NA	NA
45	2	default	S24_Malfoy	6102595	NA	Captive	MAW	M	A	Winter	Healesville	5.1041096	8.02	NA	NA	NA		NA	NA	TAACCGGT+ATCGTCTC	6,908,416	1.49	97.09	2.91	2,086	100	91.90	35.56	NA	NA	NA
46	2	default	S24_Licorice	6102596	NA	Captive	FSA	F	S	Autumn	Healesville	1.8246575	5.94	NA	NA	NA		NA	NA	TTGCAGAC+ATCGTGGT	4,475,461	0.96	95.05	4.95	1,352	100	91.42	35.46	NA	NA	NA
47	2	default	S25_Tofu	6102599	NA	Captive	FAU	F	A	Summer	Healesville	2.7452055	N/A	NA	NA	NA		NA	NA	GAGACGAT+ACCGGTTA	12,024,421	2.59	95.37	4.63	3,631	100	93.06	35.79	NA	NA	NA
48	2	default	S25_Colette	6102600	NA	Captive	FAP	F	A	Spring	Healesville	6.6273973	7.52	NA	NA	NA		NA	NA	CTTGTCGA+CGATGTTC	10,751,797	2.32	97.49	2.51	3,247	100	92.77	35.74	NA	NA	NA
49	2	default	S25_CannonBall	6102601	NA	Captive	MSW	M	S	Winter	Healesville	1.1479452	6.54	NA	NA	NA		NA	NA	TACGCTAC+CGTGTGAT	13,015,081	2.81	97.36	2.64	3,931	100	92.61	35.70	NA	NA	NA
50	2	default	S25_Cornelius	6102602	NA	Captive	MAU	M	A	Summer	Healesville	3.6904110	7.74	NA	NA	NA		NA	NA	GAATCCGA+AGCTAGTG	3,336,983	0.72	97.10	2.90	1,008	100	93.09	35.79	NA	NA	NA
51	2	default	S25_Mardook	6102603	NA	Captive	MAW	M	A	Winter	Healesville	4.3013699	9.16	NA	NA	NA		NA	NA	GCACAACT+CTGAACGT	6,826,313	1.47	97.66	2.34	2,062	100	92.67	35.71	NA	NA	NA
52	2	default	S25_Negroni	6102604	NA	Captive	FSP	F	S	Spring	Healesville	1.6767123	5.86	NA	NA	NA		NA	NA	TGTGACTG+AGCCTATC	4,566,180	0.98	97.68	2.32	1,379	100	92.13	35.61	NA	NA	NA

col <- as.character(as.numeric(grepl("Monarto",ssCLAV$Location)))
col <- gsub("0","darkblue",col)
col <- gsub("1","blue",col)

mds<-plotMDS(x1,pch=19,cex=3,col=col,main="MDS plot: Captive location: Healesville vs Monarto devils")
mtext("Healesville=Dark Blue, Monarto=Light Blue")

ssCLAV$CapLoc <- grepl("Monarto",ssCLAV$Location)
ssCLAV$MvF <- factor(ssCLAV$Sex)
ssCLAV$SubAdvAd <- as.numeric(factor(ssCLAV$Age,levels=c("S","A"),ordered = TRUE))
ssCLAV$Season <- factor(ssCLAV$Season)
str(ssCLAV)

## 'data.frame':    48 obs. of  34 variables:
##  $ Lane                   : int  1 1 1 1 1 1 1 1 1 1 ...
##  $ Project                : chr  "default" "default" "default" "default" ...
##  $ Client.ID              : chr  "S2_B99008" "S2_Bonney" "S3_Jessie" "S3_McCauley" ...
##  $ Sample                 : int  6102468 6102469 6102476 6102477 6102484 6102485 6102496 6102497 6102514 6102515 ...
##  $ X                      : logi  NA NA NA NA NA NA ...
##  $ Category               : chr  "Captive" "Captive" "Captive" "Captive" ...
##  $ Sex.Age.Season         : chr  "FAU" "FAU" "FAW" "MAA" ...
##  $ Sex                    : chr  "F" "F" "F" "M" ...
##  $ Age                    : chr  "A" "A" "A" "A" ...
##  $ Season                 : Factor w/ 4 levels "Autumn","Spring",..: 3 3 4 1 3 4 1 3 2 3 ...
##  $ Location               : chr  "Monarto" "Healesville" "Monarto" "Healesville" ...
##  $ Agebyyear              : num  3.74 2.59 2.19 3.03 3 ...
##  $ Weight                 : chr  "8.12" "5.4" "12.95" "8.52" ...
##  $ X.1                    : logi  NA NA NA NA NA NA ...
##  $ Number.of.Tumours      : int  NA NA NA NA NA NA NA NA NA NA ...
##  $ Ulcerated              : logi  NA NA NA NA NA NA ...
##  $ UlcerationDepth        : chr  "" "" "" "" ...
##  $ Secondary.infection    : logi  NA NA NA NA NA NA ...
##  $ Total.tumour.volume    : num  NA NA NA NA NA NA NA NA NA NA ...
##  $ Barcode.sequence       : chr  "GCTATCCT+AACAGGTG" "ACGTTCAG+AGTTGTGC" "CCGTAAGA+AGAGTCCA" "CAACACCT+AGCTACCA" ...
##  $ PF.Clusters            : chr  "5,338,890" "14,968,459" "4,340,498" "16,944,152" ...
##  $ X..of.the.lane         : num  1 3 1 3 1 1 1 2 1 2 ...
##  $ X..Perfect.barcode     : num  98.1 97.8 98 98 96.2 ...
##  $ X..One.mismatch.barcode: num  1.92 2.18 2.05 2.02 3.79 2.67 2.23 3.06 3.27 2.48 ...
##  $ Yield..Mbases.         : chr  "1,612" "4,520" "1,311" "5,117" ...
##  $ X..PF.Cluster          : int  100 100 100 100 100 100 100 100 100 100 ...
##  $ X.....Q30              : num  93.7 93.7 93.5 93 93.9 ...
##  $ Mean.Quality           : num  35.9 35.9 35.9 35.8 35.9 ...
##  $ X.2                    : logi  NA NA NA NA NA NA ...
##  $ X.3                    : logi  NA NA NA NA NA NA ...
##  $ X.4                    : logi  NA NA NA NA NA NA ...
##  $ CapLoc                 : logi  TRUE FALSE TRUE FALSE TRUE FALSE ...
##  $ MvF                    : Factor w/ 2 levels "F","M": 1 1 1 2 2 2 2 1 2 2 ...
##  $ SubAdvAd               : num  2 2 2 2 2 2 2 1 1 2 ...

dds <- DESeqDataSetFromMatrix(countData = x1 , colData = ssCLAV, design = ~ Season + SubAdvAd + MvF + CapLoc )

## converting counts to integer mode

##   the design formula contains one or more numeric variables with integer values,
##   specifying a model with increasing fold change for higher values.
##   did you mean for this to be a factor? if so, first convert
##   this variable to a factor using the factor() function

res <- DESeq(dds)

## estimating size factors

## estimating dispersions

## gene-wise dispersion estimates

## mean-dispersion relationship

## final dispersion estimates

## fitting model and testing

## -- replacing outliers and refitting for 10 genes
## -- DESeq argument 'minReplicatesForReplace' = 7 
## -- original counts are preserved in counts(dds)

## estimating dispersions

## fitting model and testing

z<- results(res)
vsd <- vst(dds, blind=FALSE,nsub=nrow(x1)-50)
zz<-cbind(as.data.frame(z),assay(vsd))
dge<-as.data.frame(zz[order(zz$pvalue),])
dgeCLAV <- dge
sig <- subset(dge,padj<0.05)
sig1_up <- rownames(subset(sig,log2FoldChange>0))
sig1_dn <- rownames(subset(sig,log2FoldChange<0))
length(sig1_up)

## [1] 1

maplot(dgeCLAV,"Captive loacation: Healesville vs Monarto devils")

make_volcano(dgeCLAV,"Captive loacation: Healesville vs Monarto devils")

sig[1:150,1:6] %>%
  kbl(caption="Comparison of Wild vs Captive devils") %>%
  kable_paper("hover", full_width = F)

Comparison of Wild vs Captive devils

	baseMean	log2FoldChange	lfcSE	stat	pvalue	padj
IDO1	5484.326	3.144914	0.7196113	4.370296	1.24e-05	0.0031764
NA	NA	NA	NA	NA	NA	NA
NA.1	NA	NA	NA	NA	NA	NA
NA.2	NA	NA	NA	NA	NA	NA
NA.3	NA	NA	NA	NA	NA	NA
NA.4	NA	NA	NA	NA	NA	NA
NA.5	NA	NA	NA	NA	NA	NA
NA.6	NA	NA	NA	NA	NA	NA
NA.7	NA	NA	NA	NA	NA	NA
NA.8	NA	NA	NA	NA	NA	NA
NA.9	NA	NA	NA	NA	NA	NA
NA.10	NA	NA	NA	NA	NA	NA
NA.11	NA	NA	NA	NA	NA	NA
NA.12	NA	NA	NA	NA	NA	NA
NA.13	NA	NA	NA	NA	NA	NA
NA.14	NA	NA	NA	NA	NA	NA
NA.15	NA	NA	NA	NA	NA	NA
NA.16	NA	NA	NA	NA	NA	NA
NA.17	NA	NA	NA	NA	NA	NA
NA.18	NA	NA	NA	NA	NA	NA
NA.19	NA	NA	NA	NA	NA	NA
NA.20	NA	NA	NA	NA	NA	NA
NA.21	NA	NA	NA	NA	NA	NA
NA.22	NA	NA	NA	NA	NA	NA
NA.23	NA	NA	NA	NA	NA	NA
NA.24	NA	NA	NA	NA	NA	NA
NA.25	NA	NA	NA	NA	NA	NA
NA.26	NA	NA	NA	NA	NA	NA
NA.27	NA	NA	NA	NA	NA	NA
NA.28	NA	NA	NA	NA	NA	NA
NA.29	NA	NA	NA	NA	NA	NA
NA.30	NA	NA	NA	NA	NA	NA
NA.31	NA	NA	NA	NA	NA	NA
NA.32	NA	NA	NA	NA	NA	NA
NA.33	NA	NA	NA	NA	NA	NA
NA.34	NA	NA	NA	NA	NA	NA
NA.35	NA	NA	NA	NA	NA	NA
NA.36	NA	NA	NA	NA	NA	NA
NA.37	NA	NA	NA	NA	NA	NA
NA.38	NA	NA	NA	NA	NA	NA
NA.39	NA	NA	NA	NA	NA	NA
NA.40	NA	NA	NA	NA	NA	NA
NA.41	NA	NA	NA	NA	NA	NA
NA.42	NA	NA	NA	NA	NA	NA
NA.43	NA	NA	NA	NA	NA	NA
NA.44	NA	NA	NA	NA	NA	NA
NA.45	NA	NA	NA	NA	NA	NA
NA.46	NA	NA	NA	NA	NA	NA
NA.47	NA	NA	NA	NA	NA	NA
NA.48	NA	NA	NA	NA	NA	NA
NA.49	NA	NA	NA	NA	NA	NA
NA.50	NA	NA	NA	NA	NA	NA
NA.51	NA	NA	NA	NA	NA	NA
NA.52	NA	NA	NA	NA	NA	NA
NA.53	NA	NA	NA	NA	NA	NA
NA.54	NA	NA	NA	NA	NA	NA
NA.55	NA	NA	NA	NA	NA	NA
NA.56	NA	NA	NA	NA	NA	NA
NA.57	NA	NA	NA	NA	NA	NA
NA.58	NA	NA	NA	NA	NA	NA
NA.59	NA	NA	NA	NA	NA	NA
NA.60	NA	NA	NA	NA	NA	NA
NA.61	NA	NA	NA	NA	NA	NA
NA.62	NA	NA	NA	NA	NA	NA
NA.63	NA	NA	NA	NA	NA	NA
NA.64	NA	NA	NA	NA	NA	NA
NA.65	NA	NA	NA	NA	NA	NA
NA.66	NA	NA	NA	NA	NA	NA
NA.67	NA	NA	NA	NA	NA	NA
NA.68	NA	NA	NA	NA	NA	NA
NA.69	NA	NA	NA	NA	NA	NA
NA.70	NA	NA	NA	NA	NA	NA
NA.71	NA	NA	NA	NA	NA	NA
NA.72	NA	NA	NA	NA	NA	NA
NA.73	NA	NA	NA	NA	NA	NA
NA.74	NA	NA	NA	NA	NA	NA
NA.75	NA	NA	NA	NA	NA	NA
NA.76	NA	NA	NA	NA	NA	NA
NA.77	NA	NA	NA	NA	NA	NA
NA.78	NA	NA	NA	NA	NA	NA
NA.79	NA	NA	NA	NA	NA	NA
NA.80	NA	NA	NA	NA	NA	NA
NA.81	NA	NA	NA	NA	NA	NA
NA.82	NA	NA	NA	NA	NA	NA
NA.83	NA	NA	NA	NA	NA	NA
NA.84	NA	NA	NA	NA	NA	NA
NA.85	NA	NA	NA	NA	NA	NA
NA.86	NA	NA	NA	NA	NA	NA
NA.87	NA	NA	NA	NA	NA	NA
NA.88	NA	NA	NA	NA	NA	NA
NA.89	NA	NA	NA	NA	NA	NA
NA.90	NA	NA	NA	NA	NA	NA
NA.91	NA	NA	NA	NA	NA	NA
NA.92	NA	NA	NA	NA	NA	NA
NA.93	NA	NA	NA	NA	NA	NA
NA.94	NA	NA	NA	NA	NA	NA
NA.95	NA	NA	NA	NA	NA	NA
NA.96	NA	NA	NA	NA	NA	NA
NA.97	NA	NA	NA	NA	NA	NA
NA.98	NA	NA	NA	NA	NA	NA
NA.99	NA	NA	NA	NA	NA	NA
NA.100	NA	NA	NA	NA	NA	NA
NA.101	NA	NA	NA	NA	NA	NA
NA.102	NA	NA	NA	NA	NA	NA
NA.103	NA	NA	NA	NA	NA	NA
NA.104	NA	NA	NA	NA	NA	NA
NA.105	NA	NA	NA	NA	NA	NA
NA.106	NA	NA	NA	NA	NA	NA
NA.107	NA	NA	NA	NA	NA	NA
NA.108	NA	NA	NA	NA	NA	NA
NA.109	NA	NA	NA	NA	NA	NA
NA.110	NA	NA	NA	NA	NA	NA
NA.111	NA	NA	NA	NA	NA	NA
NA.112	NA	NA	NA	NA	NA	NA
NA.113	NA	NA	NA	NA	NA	NA
NA.114	NA	NA	NA	NA	NA	NA
NA.115	NA	NA	NA	NA	NA	NA
NA.116	NA	NA	NA	NA	NA	NA
NA.117	NA	NA	NA	NA	NA	NA
NA.118	NA	NA	NA	NA	NA	NA
NA.119	NA	NA	NA	NA	NA	NA
NA.120	NA	NA	NA	NA	NA	NA
NA.121	NA	NA	NA	NA	NA	NA
NA.122	NA	NA	NA	NA	NA	NA
NA.123	NA	NA	NA	NA	NA	NA
NA.124	NA	NA	NA	NA	NA	NA
NA.125	NA	NA	NA	NA	NA	NA
NA.126	NA	NA	NA	NA	NA	NA
NA.127	NA	NA	NA	NA	NA	NA
NA.128	NA	NA	NA	NA	NA	NA
NA.129	NA	NA	NA	NA	NA	NA
NA.130	NA	NA	NA	NA	NA	NA
NA.131	NA	NA	NA	NA	NA	NA
NA.132	NA	NA	NA	NA	NA	NA
NA.133	NA	NA	NA	NA	NA	NA
NA.134	NA	NA	NA	NA	NA	NA
NA.135	NA	NA	NA	NA	NA	NA
NA.136	NA	NA	NA	NA	NA	NA
NA.137	NA	NA	NA	NA	NA	NA
NA.138	NA	NA	NA	NA	NA	NA
NA.139	NA	NA	NA	NA	NA	NA
NA.140	NA	NA	NA	NA	NA	NA
NA.141	NA	NA	NA	NA	NA	NA
NA.142	NA	NA	NA	NA	NA	NA
NA.143	NA	NA	NA	NA	NA	NA
NA.144	NA	NA	NA	NA	NA	NA
NA.145	NA	NA	NA	NA	NA	NA
NA.146	NA	NA	NA	NA	NA	NA
NA.147	NA	NA	NA	NA	NA	NA
NA.148	NA	NA	NA	NA	NA	NA

write.table(dge,file="dgeWvCAV.tsv",sep="\t")

ssCvD <- filter(ss, Category %in% c('DFT1','DFT2', 'Captive'))

ssCvDAV <- ssCvD[which(grepl("Captive",ssCvD$Category) + grepl("DFT",ssCvD$Category)>0),]
ssCvDAV

##    Lane Project              Client.ID  Sample  X Category Sex.Age.Season Sex
## 1     1 default            S1_Sarafina 6102463 NA     DFT1            FAP   F
## 2     1 default                S1_Lucy 6102464 NA     DFT1            FAU   F
## 3     1 default              S2_B99008 6102468 NA  Captive            FAU   F
## 4     1 default              S2_Bonney 6102469 NA  Captive            FAU   F
## 5     1 default                S2_Hope 6102471 NA     DFT1            FAU   F
## 6     1 default               S2_Sally 6102472 NA     DFT1            FAU   F
## 7     1 default                 S3_Apu 6102473 NA     DFT1            MAA   M
## 8     1 default              S3_Simone 6102474 NA     DFT1            FAW   F
## 9     1 default              S3_Jessie 6102476 NA  Captive            FAW   F
## 10    1 default            S3_McCauley 6102477 NA  Captive            MAA   M
## 11    1 default              S4_Wiggum 6102480 NA     DFT1            MAA   M
## 12    1 default                S4_Pooh 6102482 NA     DFT1            MAA   M
## 13    1 default            S4_Milhouse 6102483 NA     DFT1            MAA   M
## 14    1 default              S5_Fenton 6102484 NA  Captive            MAU   M
## 15    1 default         S5_Bryan Adams 6102485 NA  Captive            MAW   M
## 16    1 default               S5_Allen 6102488 NA     DFT1            MAU   M
## 17    1 default                S5_Cyan 6102489 NA     DFT1            MAW   M
## 18    1 default              S6_Shelob 6102492 NA     DFT2            FAP   F
## 19    1 default           S6_Stuttgart 6102493 NA     DFT1            MSP   M
## 20    1 default               S7_Fredo 6102496 NA  Captive            MAA   M
## 21    1 default              S7_Piglet 6102497 NA  Captive            FSU   F
## 22    1 default              S7_Violet 6102498 NA     DFT2            FSU   F
## 23    1 default              S7_Willie 6102500 NA     DFT2            MAA   M
## 24    1 default              S8_Orange 6102502 NA     DFT1            FAU   F
## 25    1 default                S8_ST67 6102505 NA     DFT2            MAW   M
## 26    1 default              S8_Yellow 6102507 NA     DFT2            MAU   M
## 27    1 default                S9_Jane 6102508 NA     DFT1            FAA   F
## 28    1 default        S9_Clytemnestra 6102511 NA     DFT1            FSP   F
## 29    1 default                S9_Zeus 6102512 NA     DFT2            MSW   M
## 30    1 default            S10_McMahon 6102514 NA  Captive            MSP   M
## 31    1 default              S10_Logan 6102515 NA  Captive            MAU   M
## 32    1 default           S10_Laurasia 6102516 NA     DFT2            MSP   M
## 33    1 default          S10_Enchilada 6102517 NA     DFT2            MAP   M
## 34    1 default S11_Murdoch/Washington 6102519 NA     DFT2            MAA   M
## 35    1 default               S11_Xuki 6102520 NA  Captive            MAA   M
## 36    1 default           S11_Kalamata 6102521 NA  Captive            MSW   M
## 37    1 default                S12_Abe 6102524 NA     DFT1            MAA   M
## 38    1 default              S12_Lilac 6102526 NA     DFT1            MAW   M
## 39    1 default              S12_Wally 6102528 NA  Captive            MAA   M
## 40    1 default                S12_Naz 6102529 NA  Captive            MAW   M
## 41    1 default             S13_Humbug 6102530 NA  Captive            MAU   M
## 42    1 default              S13_Green 6102531 NA     DFT1            MAA   M
## 43    1 default               S13_Bump 6102532 NA     DFT1            MAU   M
## 44    1 default             S13_Xerxes 6102535 NA  Captive            MAA   M
## 45    1 default             S14_Purple 6102536 NA     DFT1            FAW   F
## 46    1 default             S14_Catani 6102538 NA  Captive            FAU   F
## 47    2 default             S15_Amelia 6102541 NA  Captive            FAU   F
## 48    2 default             S15_London 6102542 NA  Captive            FAU   F
## 49    2 default              S15_Elyse 6102543 NA  Captive            FAP   F
## 50    2 default             S16_Marley 6102547 NA  Captive            FAU   F
## 51    2 default              S16_Milla 6102548 NA  Captive            FAU   F
## 52    2 default         S16_Maleficent 6102549 NA  Captive            FAU   F
## 53    2 default             S17_Anirak 6102553 NA  Captive            FSW   F
## 54    2 default               S17_Poly 6102554 NA  Captive            FSU   F
## 55    2 default             S17_Methyl 6102556 NA  Captive            FSU   F
## 56    2 default               S18_Zara 6102558 NA  Captive            FSW   F
## 57    2 default               S18_Maze 6102559 NA  Captive            FSW   F
## 58    2 default            S18_Delores 6102560 NA  Captive            FSW   F
## 59    2 default            S19_Morocco 6102564 NA  Captive            MAU   M
## 60    2 default           S19_Pecorino 6102565 NA  Captive            FSW   F
## 61    2 default              S19_Hooch 6102566 NA  Captive            MSA   M
## 62    2 default              S20_Muasy 6102569 NA  Captive            MSU   M
## 63    2 default               S20_Fuse 6102570 NA  Captive            MSU   M
## 64    2 default             S20_Joseph 6102571 NA  Captive            MSU   M
## 65    2 default              S21_Moama 6102576 NA  Captive            MAW   M
## 66    2 default             S21_Nauset 6102577 NA  Captive            MSW   M
## 67    2 default             S21_Graham 6102578 NA  Captive            MSW   M
## 68    2 default              S22_Yanni 6102581 NA  Captive            MSW   M
## 69    2 default            S22_Gherkin 6102582 NA  Captive            MSW   M
## 70    2 default            S22_Stilton 6102584 NA  Captive            MSW   M
## 71    2 default             S23_Osiris 6102588 NA  Captive            MSW   M
## 72    2 default             S23_X-boxx 6102589 NA  Captive            FSW   F
## 73    2 default             S23_Quince 6102590 NA  Captive            MSW   M
## 74    2 default             S24_Malfoy 6102595 NA  Captive            MAW   M
## 75    2 default           S24_Licorice 6102596 NA  Captive            FSA   F
## 76    2 default               S25_Tofu 6102599 NA  Captive            FAU   F
## 77    2 default            S25_Colette 6102600 NA  Captive            FAP   F
## 78    2 default         S25_CannonBall 6102601 NA  Captive            MSW   M
## 79    2 default          S25_Cornelius 6102602 NA  Captive            MAU   M
## 80    2 default            S25_Mardook 6102603 NA  Captive            MAW   M
## 81    2 default            S25_Negroni 6102604 NA  Captive            FSP   F
##    Age Season    Location Agebyyear Weight X.1 Number.of.Tumours Ulcerated
## 1    A Spring     Channel 2.6191781    6.4  NA                NA        NA
## 2    A Summer     Channel 2.9178082    7.1  NA                 3      TRUE
## 3    A Summer     Monarto 3.7369863   8.12  NA                NA        NA
## 4    A Summer Healesville 2.5863014    5.4  NA                NA        NA
## 5    A Summer     Channel 2.9315068    6.4  NA                 4      TRUE
## 6    A Summer     Channel 2.9945205      7  NA                 1      TRUE
## 7    A Autumn     Channel 3.1150685    9.5  NA                 1     FALSE
## 8    A Winter     Channel 3.2739726   10.2  NA                 1      TRUE
## 9    A Winter     Monarto 2.1945205  12.95  NA                NA        NA
## 10   A Autumn Healesville 3.0328767   8.52  NA                NA        NA
## 11   A Autumn     Channel 3.1178082    8.3  NA                 2     FALSE
## 12   A Autumn     Channel 4.1863014   11.8  NA                 1     FALSE
## 13   A Autumn     Channel 4.1013699     11  NA                 1      TRUE
## 14   A Summer     Monarto 3.0000000    N/A  NA                NA        NA
## 15   A Winter Healesville 4.1616438    8.6  NA                NA        NA
## 16   A Summer     Channel 1.9095890    6.7  NA                 1      TRUE
## 17   A Winter     Channel 3.3945205    9.1  NA                 5      TRUE
## 18   A Spring     Channel 4.6219178    7.2  NA                NA        NA
## 19   S Spring     Channel 0.5780822   12.7  NA                 1     FALSE
## 20   A Autumn Healesville 2.9945205    8.8  NA                NA        NA
## 21   S Summer Healesville 1.5890411   4.76  NA                NA        NA
## 22   S Summer     Channel 0.8739726    4.6  NA                 1      TRUE
## 23   A Autumn     Channel 2.1150685    8.2  NA                 2      TRUE
## 24   A Summer     Channel 1.9150685    7.1  NA                 7      TRUE
## 25   A Winter     Channel 4.2794521   10.2  NA                 1      TRUE
## 26   A Summer     Channel 1.8657534   10.5  NA                 2      TRUE
## 27   A Autumn     Channel 3.1972603    6.1  NA                NA        NA
## 28   S Spring     Channel 1.6602740      6  NA                NA        NA
## 29   S Winter     Channel 1.3479452    7.3  NA                NA        NA
## 30   S Spring Healesville 1.6493151   8.06  NA                NA        NA
## 31   A Summer Healesville 5.6000000   9.02  NA                NA        NA
## 32   S Spring     Channel 1.6383562      8  NA                NA        NA
## 33   A Spring     Channel 3.6438356    9.3  NA                14     FALSE
## 34   A Autumn     Channel 1.4438356    8.9  NA                NA        NA
## 35   A Autumn     Monarto 2.0767123   7.95  NA                NA        NA
## 36   S Winter Healesville 1.2849315   5.58  NA                NA        NA
## 37   A Autumn     Channel 3.1150685    7.9  NA                 1      TRUE
## 38   A Winter     Channel 3.4109589    9.6  NA                 5      TRUE
## 39   A Autumn     Monarto 2.1616438    N/A  NA                NA        NA
## 40   A Winter     Monarto 4.4547945    N/A  NA                NA        NA
## 41   A Summer     Monarto 2.5397260   7.94  NA                NA        NA
## 42   A Autumn     Channel 3.1534247    7.5  NA                 2      TRUE
## 43   A Summer     Channel 3.9068493    9.3  NA                NA        NA
## 44   A Autumn     Monarto 2.0767123    N/A  NA                NA        NA
## 45   A Winter     Channel 2.3945205    6.8  NA                 1      TRUE
## 46   A Summer Healesville 2.8630137   6.46  NA                NA        NA
## 47   A Summer Healesville 4.7260274   5.85  NA                NA        NA
## 48   A Summer Healesville 3.7863014    4.8  NA                NA        NA
## 49   A Spring Healesville 4.5232877   7.88  NA                NA        NA
## 50   A Summer     Monarto 4.7726027   7.51  NA                NA        NA
## 51   A Summer Healesville 2.7671233   5.82  NA                NA        NA
## 52   A Summer     Monarto 4.5589041   8.59  NA                NA        NA
## 53   S Winter Healesville 1.3424658   5.42  NA                NA        NA
## 54   S Summer     Monarto 1.7917808   6.67  NA                NA        NA
## 55   S Summer Healesville 0.7534247   2.72  NA                NA        NA
## 56   S Winter     Monarto 1.5150685    4.6  NA                NA        NA
## 57   S Winter Healesville 1.4000000   5.82  NA                NA        NA
## 58   S Winter Healesville 1.2986301    5.1  NA                NA        NA
## 59   A Summer     Monarto 4.5917808   9.82  NA                NA        NA
## 60   S Winter Healesville 1.1123288   6.02  NA                NA        NA
## 61   S Autumn Healesville 1.0164384    6.1  NA                NA        NA
## 62   S Summer Healesville 1.4219178   6.62  NA                NA        NA
## 63   S Summer Healesville 1.7315068   6.26  NA                NA        NA
## 64   S Summer Healesville 1.7479452   7.42  NA                NA        NA
## 65   A Winter     Monarto 1.6986301   8.75  NA                NA        NA
## 66   S Winter Healesville 1.3041096  4.036  NA                NA        NA
## 67   S Winter Healesville 1.3616438    6.4  NA                NA        NA
## 68   S Winter     Monarto 1.5178082   8.15  NA                NA        NA
## 69   S Winter Healesville 1.2630137   6.32  NA                NA        NA
## 70   S Winter Healesville 1.2986301   5.76  NA                NA        NA
## 71   S Winter Healesville 1.1753425   5.34  NA                NA        NA
## 72   S Winter     Monarto 1.4657534   4.34  NA                NA        NA
## 73   S Winter Healesville 1.3835616    N/A  NA                NA        NA
## 74   A Winter Healesville 5.1041096   8.02  NA                NA        NA
## 75   S Autumn Healesville 1.8246575   5.94  NA                NA        NA
## 76   A Summer Healesville 2.7452055    N/A  NA                NA        NA
## 77   A Spring Healesville 6.6273973   7.52  NA                NA        NA
## 78   S Winter Healesville 1.1479452   6.54  NA                NA        NA
## 79   A Summer Healesville 3.6904110   7.74  NA                NA        NA
## 80   A Winter Healesville 4.3013699   9.16  NA                NA        NA
## 81   S Spring Healesville 1.6767123   5.86  NA                NA        NA
##    UlcerationDepth Secondary.infection Total.tumour.volume  Barcode.sequence
## 1                                   NA           95140.790 ATCCAGAG+TAACGTCG
## 2             deep                TRUE          700825.170 AAGCACTG+AGGTCAAC
## 3                                   NA                  NA GCTATCCT+AACAGGTG
## 4                                   NA                  NA ACGTTCAG+AGTTGTGC
## 5             deep                TRUE          103815.530 AGCACTTC+ACGTCCAA
## 6             deep               FALSE            1588.272 CGTTGCAA+ATAGAGCG
## 7      superficial                TRUE             200.000 CCTTGATC+GATGGAGT
## 8      superficial                TRUE            2670.678 CGGTTGTT+GTGGTATG
## 9                                   NA                  NA CCGTAAGA+AGAGTCCA
## 10                                  NA                  NA CAACACCT+AGCTACCA
## 11                               FALSE            7272.309 GTGAAGTG+GAGCAATC
## 12                               FALSE            7577.856 CTTCGACT+GATCTTGC
## 13            deep                TRUE             480.000 AACCGTTC+CTAGCTCA
## 14                                  NA                  NA GATTACCG+TTGGACTG
## 15                                  NA                  NA CACTAGCT+TCGGATTC
## 16            deep                TRUE           55959.750 TTGCAGAC+ATCGTGGT
## 17            deep                TRUE          206006.422 GTAGGAGT+GTCCTAAG
## 18                                  NA            7455.435 GTCATCGA+GATCCACT
## 19                                TRUE            3809.806 AGCGTGTT+CTTCGCAA
## 20                                  NA                  NA AGTGGATC+TCGATGAC
## 21                                  NA                  NA ATCACACG+GTAGCGTA
## 22     superficial                TRUE            5517.330 CTTGTCGA+CGATGTTC
## 23            deep                TRUE            1527.480 TGGACTCT+TCTTACGG
## 24     superficial                TRUE           43690.461 GTTGACCT+CAGTGCTT
## 25            deep                TRUE          102246.432 TTGGACGT+GAAGTGCT
## 26            deep                TRUE          139389.671 GAAGTTGG+CGAACAAC
## 27                                  NA           64646.036 GTTCTCGT+TGTTCCGT
## 28                                  NA           34738.080 CATGGCTA+CACACATC
## 29                                  NA             100.000 CGTGTGTA+CCAACGAA
## 30                                  NA                  NA AACGTCTG+AATGACGC
## 31                                  NA                  NA GTGCCATA+CTCCTAGT
## 32                                  NA            2697.152 TACGCTAC+CGTGTGAT
## 33                               FALSE          301301.905 GATTGCTC+CACTTCAC
## 34                                  NA          127176.305 ACTGAGGT+GTCAACAG
## 35                                  NA                  NA ATCGCCAT+TGCGTAAC
## 36                                  NA                  NA GAACATCG+TCGACAAG
## 37            deep                TRUE           13125.000 GTTGTTCG+CCAACTTC
## 38     superficial                TRUE           37293.844 TATCGGTC+TACCGGAT
## 39                                  NA                  NA ACGACTTG+AACAGCGA
## 40                                  NA                  NA CCTTGTAG+CCTTGGAA
## 41                                  NA                  NA GAGCAGTA+TGAGCTGT
## 42     superficial                TRUE            7407.222 GTCGAAGA+CCACATTG
## 43                                  NA          629522.362 AGTTCGTC+ACTCGATC
## 44                                  NA                  NA GATAGGCT+CAGTCACA
## 45            deep                TRUE           24938.123 CGCTCTAT+TTGCAACG
## 46                                  NA                  NA TGAAGACG+ACATGCCA
## 47                                  NA                  NA AACGTCTG+AATGACGC
## 48                                  NA                  NA GTCATCGA+GATCCACT
## 49                                  NA                  NA AGCGTGTT+CTTCGCAA
## 50                                  NA                  NA GCAAGATC+AGTCGAAG
## 51                                  NA                  NA TTGCGAAG+AACACGCT
## 52                                  NA                  NA TGCTTCCA+CGATCGAT
## 53                                  NA                  NA TGAGCTAG+GAACGGTT
## 54                                  NA                  NA CTTAGGAC+ACTCCTAC
## 55                                  NA                  NA TGGTACAG+TCGAGAGT
## 56                                  NA                  NA ACCACGAT+GTCTGCAA
## 57                                  NA                  NA ACTGAGGT+GTCAACAG
## 58                                  NA                  NA ACGACTTG+AACAGCGA
## 59                                  NA                  NA TCAGGCTT+ATCATGCG
## 60                                  NA                  NA ATGCCTGT+AGATTGCG
## 61                                  NA                  NA CCGTAAGA+AGAGTCCA
## 62                                  NA                  NA TTCCAAGG+CTACAAGG
## 63                                  NA                  NA GTAGGAGT+GTCCTAAG
## 64                                  NA                  NA CTTCGACT+GATCTTGC
## 65                                  NA                  NA TTGGACGT+GAAGTGCT
## 66                                  NA                  NA TATCGGTC+TACCGGAT
## 67                                  NA                  NA TACGCCTT+GCTACTCT
## 68                                  NA                  NA CGTGTGTA+CCAACGAA
## 69                                  NA                  NA CCTTGTAG+CCTTGGAA
## 70                                  NA                  NA CGCAATCT+ACAGGCAT
## 71                                  NA                  NA AGTGGATC+TCGATGAC
## 72                                  NA                  NA CTGTTGAC+ACCTCAGT
## 73                                  NA                  NA AACCGTTC+CTAGCTCA
## 74                                  NA                  NA TAACCGGT+ATCGTCTC
## 75                                  NA                  NA TTGCAGAC+ATCGTGGT
## 76                                  NA                  NA GAGACGAT+ACCGGTTA
## 77                                  NA                  NA CTTGTCGA+CGATGTTC
## 78                                  NA                  NA TACGCTAC+CGTGTGAT
## 79                                  NA                  NA GAATCCGA+AGCTAGTG
## 80                                  NA                  NA GCACAACT+CTGAACGT
## 81                                  NA                  NA TGTGACTG+AGCCTATC
##    PF.Clusters X..of.the.lane X..Perfect.barcode X..One.mismatch.barcode
## 1    6,491,488           1.00              74.07                   25.93
## 2    6,121,785           1.00              98.00                    2.00
## 3    5,338,890           1.00              98.08                    1.92
## 4   14,968,459           3.00              97.82                    2.18
## 5    1,971,776           0.00              97.96                    2.04
## 6    2,429,449           1.00              97.42                    2.58
## 7    4,939,580           1.00              96.94                    3.06
## 8    4,015,027           1.00              95.91                    4.09
## 9    4,340,498           1.00              97.95                    2.05
## 10  16,944,152           3.00              97.98                    2.02
## 11   4,201,355           1.00              97.17                    2.83
## 12   3,821,151           1.00              97.31                    2.69
## 13  10,492,900           2.00              98.13                    1.87
## 14   5,125,566           1.00              96.21                    3.79
## 15   5,317,767           1.00              97.33                    2.67
## 16   5,288,435           1.00              96.85                    3.15
## 17   4,138,600           1.00              98.03                    1.97
## 18   3,184,798           1.00              98.21                    1.79
## 19   3,056,475           1.00              97.50                    2.50
## 20   2,690,632           1.00              97.77                    2.23
## 21  12,134,750           2.00              96.94                    3.06
## 22   2,768,682           1.00              98.19                    1.81
## 23   4,539,036           1.00              98.30                    1.70
## 24   5,072,309           1.00              97.92                    2.08
## 25   5,476,735           1.00              95.44                    4.56
## 26   6,469,324           1.00              97.99                    2.01
## 27  12,489,105           3.00              97.92                    2.08
## 28   6,141,790           1.00              98.00                    2.00
## 29     699,953           0.00              97.64                    2.36
## 30   3,809,607           1.00              96.73                    3.27
## 31   9,606,720           2.00              97.52                    2.48
## 32   9,716,148           2.00              97.81                    2.19
## 33  11,967,844           2.00              97.82                    2.18
## 34   2,003,509           0.00              97.82                    2.18
## 35   6,018,884           1.00              97.88                    2.12
## 36   6,861,544           1.00              97.76                    2.24
## 37   6,446,467           1.00              97.60                    2.40
## 38   4,490,241           1.00              98.15                    1.85
## 39   3,000,501           1.00              97.96                    2.04
## 40   5,927,942           1.00              97.40                    2.60
## 41   1,702,608           0.00              98.07                    1.93
## 42   2,623,695           1.00              97.38                    2.62
## 43   9,435,053           2.00              98.16                    1.84
## 44   7,628,257           2.00              98.34                    1.66
## 45   5,432,182           1.00              97.81                    2.19
## 46  13,381,058           3.00              96.98                    3.02
## 47   8,041,818           1.73              96.32                    3.68
## 48   5,758,111           1.24              96.72                    3.28
## 49   5,774,637           1.24              96.76                    3.24
## 50     348,621           0.08              97.24                    2.76
## 51   6,376,997           1.37              97.60                    2.40
## 52   9,181,668           1.98              98.08                    1.92
## 53   3,740,878           0.81              97.15                    2.85
## 54  11,133,566           2.40              97.85                    2.15
## 55   7,887,375           1.70              97.57                    2.43
## 56   7,308,907           1.58              97.70                    2.30
## 57   2,984,930           0.64              97.59                    2.41
## 58          98           0.00              94.90                    5.10
## 59   4,879,629           1.05              97.38                    2.62
## 60   4,857,395           1.05              97.58                    2.42
## 61   1,074,345           0.23              97.56                    2.44
## 62  10,669,237           2.30              97.64                    2.36
## 63  10,603,179           2.29              97.34                    2.66
## 64   5,642,025           1.22              96.37                    3.63
## 65  10,821,895           2.33              96.02                    3.98
## 66   8,693,920           1.87              98.13                    1.87
## 67   3,422,010           0.74              97.72                    2.28
## 68   4,811,453           1.04              97.67                    2.33
## 69   5,913,586           1.27              97.05                    2.95
## 70   8,306,537           1.79              96.94                    3.06
## 71   3,528,471           0.76              97.24                    2.76
## 72          23           0.00             100.00                     NaN
## 73   7,247,793           1.56              97.63                    2.37
## 74   6,908,416           1.49              97.09                    2.91
## 75   4,475,461           0.96              95.05                    4.95
## 76  12,024,421           2.59              95.37                    4.63
## 77  10,751,797           2.32              97.49                    2.51
## 78  13,015,081           2.81              97.36                    2.64
## 79   3,336,983           0.72              97.10                    2.90
## 80   6,826,313           1.47              97.66                    2.34
## 81   4,566,180           0.98              97.68                    2.32
##    Yield..Mbases. X..PF.Cluster X.....Q30 Mean.Quality X.2 X.3 X.4
## 1           1,960           100     93.74        35.92  NA  NA  NA
## 2           1,849           100     93.84        35.93  NA  NA  NA
## 3           1,612           100     93.69        35.91  NA  NA  NA
## 4           4,520           100     93.68        35.91  NA  NA  NA
## 5             595           100     93.62        35.88  NA  NA  NA
## 6             734           100     93.78        35.93  NA  NA  NA
## 7           1,492           100     93.71        35.91  NA  NA  NA
## 8           1,213           100     93.42        35.86  NA  NA  NA
## 9           1,311           100     93.54        35.88  NA  NA  NA
## 10          5,117           100     93.05        35.78  NA  NA  NA
## 11          1,269           100     93.64        35.90  NA  NA  NA
## 12          1,154           100     93.60        35.89  NA  NA  NA
## 13          3,169           100     93.80        35.92  NA  NA  NA
## 14          1,548           100     93.89        35.94  NA  NA  NA
## 15          1,606           100     93.24        35.83  NA  NA  NA
## 16          1,597           100     93.42        35.86  NA  NA  NA
## 17          1,250           100     93.67        35.90  NA  NA  NA
## 18            962           100     93.90        35.94  NA  NA  NA
## 19            923           100     93.95        35.95  NA  NA  NA
## 20            813           100     93.77        35.93  NA  NA  NA
## 21          3,665           100     93.34        35.84  NA  NA  NA
## 22            836           100     93.85        35.93  NA  NA  NA
## 23          1,371           100     93.84        35.93  NA  NA  NA
## 24          1,532           100     93.36        35.84  NA  NA  NA
## 25          1,654           100     93.21        35.82  NA  NA  NA
## 26          1,954           100     94.00        35.97  NA  NA  NA
## 27          3,772           100     93.54        35.88  NA  NA  NA
## 28          1,855           100     93.42        35.85  NA  NA  NA
## 29            211           100     93.33        35.84  NA  NA  NA
## 30          1,151           100     93.07        35.79  NA  NA  NA
## 31          2,901           100     93.25        35.83  NA  NA  NA
## 32          2,934           100     93.97        35.95  NA  NA  NA
## 33          3,614           100     93.88        35.94  NA  NA  NA
## 34            605           100     93.39        35.85  NA  NA  NA
## 35          1,818           100     93.58        35.89  NA  NA  NA
## 36          2,072           100     93.49        35.87  NA  NA  NA
## 37          1,947           100     93.65        35.90  NA  NA  NA
## 38          1,356           100     93.42        35.85  NA  NA  NA
## 39            906           100     93.70        35.91  NA  NA  NA
## 40          1,790           100     93.86        35.94  NA  NA  NA
## 41            514           100     93.58        35.89  NA  NA  NA
## 42            792           100     92.94        35.76  NA  NA  NA
## 43          2,849           100     93.64        35.89  NA  NA  NA
## 44          2,304           100     93.62        35.89  NA  NA  NA
## 45          1,641           100     93.93        35.95  NA  NA  NA
## 46          4,041           100     93.36        35.84  NA  NA  NA
## 47          2,429           100     92.18        35.63  NA  NA  NA
## 48          1,739           100     91.87        35.55  NA  NA  NA
## 49          1,744           100     91.56        35.50  NA  NA  NA
## 50            105           100     92.77        35.73  NA  NA  NA
## 51          1,926           100     92.44        35.67  NA  NA  NA
## 52          2,773           100     93.04        35.79  NA  NA  NA
## 53          1,130           100     93.03        35.79  NA  NA  NA
## 54          3,362           100     92.47        35.68  NA  NA  NA
## 55          2,382           100     93.15        35.81  NA  NA  NA
## 56          2,207           100     92.84        35.75  NA  NA  NA
## 57            901           100     92.49        35.69  NA  NA  NA
## 58              0           100     89.11        34.91  NA  NA  NA
## 59          1,474           100     92.63        35.71  NA  NA  NA
## 60          1,467           100     92.64        35.72  NA  NA  NA
## 61            324           100     92.66        35.72  NA  NA  NA
## 62          3,222           100     92.79        35.74  NA  NA  NA
## 63          3,202           100     92.01        35.58  NA  NA  NA
## 64          1,704           100     91.87        35.56  NA  NA  NA
## 65          3,268           100     92.72        35.73  NA  NA  NA
## 66          2,626           100     92.32        35.65  NA  NA  NA
## 67          1,033           100     92.94        35.76  NA  NA  NA
## 68          1,453           100     92.30        35.65  NA  NA  NA
## 69          1,786           100     92.92        35.77  NA  NA  NA
## 70          2,509           100     91.86        35.55  NA  NA  NA
## 71          1,066           100     93.08        35.79  NA  NA  NA
## 72              0           100     95.85        36.37  NA  NA  NA
## 73          2,189           100     92.28        35.63  NA  NA  NA
## 74          2,086           100     91.90        35.56  NA  NA  NA
## 75          1,352           100     91.42        35.46  NA  NA  NA
## 76          3,631           100     93.06        35.79  NA  NA  NA
## 77          3,247           100     92.77        35.74  NA  NA  NA
## 78          3,931           100     92.61        35.70  NA  NA  NA
## 79          1,008           100     93.09        35.79  NA  NA  NA
## 80          2,062           100     92.67        35.71  NA  NA  NA
## 81          1,379           100     92.13        35.61  NA  NA  NA

ssCvDAV <- ssCvDAV[ssCvDAV$Sample %in% colnames(x),]

dim(ssCvDAV)

## [1] 76 31

x1 <- x[,which(colnames(x) %in% ssCvDAV$Sample)]
dim(x1)

## [1] 310  76

x1 <- x1[which(rowMeans(x1)>10),]
dim(x1)

## [1] 264  76

ssCvDAV %>%  kbl(caption="Sample sheet")%>% kable_styling("hover",full_width=FALSE) 

Sample sheet

	Lane	Project	Client.ID	Sample	X	Category	Sex.Age.Season	Sex	Age	Season	Location	Agebyyear	Weight	X.1	Number.of.Tumours	Ulcerated	UlcerationDepth	Secondary.infection	Total.tumour.volume	Barcode.sequence	PF.Clusters	X..of.the.lane	X..Perfect.barcode	X..One.mismatch.barcode	Yield..Mbases.	X..PF.Cluster	X…..Q30	Mean.Quality	X.2	X.3	X.4
1	1	default	S1_Sarafina	6102463	NA	DFT1	FAP	F	A	Spring	Channel	2.6191781	6.4	NA	NA	NA		NA	95140.790	ATCCAGAG+TAACGTCG	6,491,488	1.00	74.07	25.93	1,960	100	93.74	35.92	NA	NA	NA
2	1	default	S1_Lucy	6102464	NA	DFT1	FAU	F	A	Summer	Channel	2.9178082	7.1	NA	3	TRUE	deep	TRUE	700825.170	AAGCACTG+AGGTCAAC	6,121,785	1.00	98.00	2.00	1,849	100	93.84	35.93	NA	NA	NA
3	1	default	S2_B99008	6102468	NA	Captive	FAU	F	A	Summer	Monarto	3.7369863	8.12	NA	NA	NA		NA	NA	GCTATCCT+AACAGGTG	5,338,890	1.00	98.08	1.92	1,612	100	93.69	35.91	NA	NA	NA
4	1	default	S2_Bonney	6102469	NA	Captive	FAU	F	A	Summer	Healesville	2.5863014	5.4	NA	NA	NA		NA	NA	ACGTTCAG+AGTTGTGC	14,968,459	3.00	97.82	2.18	4,520	100	93.68	35.91	NA	NA	NA
5	1	default	S2_Hope	6102471	NA	DFT1	FAU	F	A	Summer	Channel	2.9315068	6.4	NA	4	TRUE	deep	TRUE	103815.530	AGCACTTC+ACGTCCAA	1,971,776	0.00	97.96	2.04	595	100	93.62	35.88	NA	NA	NA
6	1	default	S2_Sally	6102472	NA	DFT1	FAU	F	A	Summer	Channel	2.9945205	7	NA	1	TRUE	deep	FALSE	1588.272	CGTTGCAA+ATAGAGCG	2,429,449	1.00	97.42	2.58	734	100	93.78	35.93	NA	NA	NA
7	1	default	S3_Apu	6102473	NA	DFT1	MAA	M	A	Autumn	Channel	3.1150685	9.5	NA	1	FALSE	superficial	TRUE	200.000	CCTTGATC+GATGGAGT	4,939,580	1.00	96.94	3.06	1,492	100	93.71	35.91	NA	NA	NA
8	1	default	S3_Simone	6102474	NA	DFT1	FAW	F	A	Winter	Channel	3.2739726	10.2	NA	1	TRUE	superficial	TRUE	2670.678	CGGTTGTT+GTGGTATG	4,015,027	1.00	95.91	4.09	1,213	100	93.42	35.86	NA	NA	NA
9	1	default	S3_Jessie	6102476	NA	Captive	FAW	F	A	Winter	Monarto	2.1945205	12.95	NA	NA	NA		NA	NA	CCGTAAGA+AGAGTCCA	4,340,498	1.00	97.95	2.05	1,311	100	93.54	35.88	NA	NA	NA
10	1	default	S3_McCauley	6102477	NA	Captive	MAA	M	A	Autumn	Healesville	3.0328767	8.52	NA	NA	NA		NA	NA	CAACACCT+AGCTACCA	16,944,152	3.00	97.98	2.02	5,117	100	93.05	35.78	NA	NA	NA
11	1	default	S4_Wiggum	6102480	NA	DFT1	MAA	M	A	Autumn	Channel	3.1178082	8.3	NA	2	FALSE		FALSE	7272.309	GTGAAGTG+GAGCAATC	4,201,355	1.00	97.17	2.83	1,269	100	93.64	35.90	NA	NA	NA
12	1	default	S4_Pooh	6102482	NA	DFT1	MAA	M	A	Autumn	Channel	4.1863014	11.8	NA	1	FALSE		FALSE	7577.856	CTTCGACT+GATCTTGC	3,821,151	1.00	97.31	2.69	1,154	100	93.60	35.89	NA	NA	NA
13	1	default	S4_Milhouse	6102483	NA	DFT1	MAA	M	A	Autumn	Channel	4.1013699	11	NA	1	TRUE	deep	TRUE	480.000	AACCGTTC+CTAGCTCA	10,492,900	2.00	98.13	1.87	3,169	100	93.80	35.92	NA	NA	NA
14	1	default	S5_Fenton	6102484	NA	Captive	MAU	M	A	Summer	Monarto	3.0000000	N/A	NA	NA	NA		NA	NA	GATTACCG+TTGGACTG	5,125,566	1.00	96.21	3.79	1,548	100	93.89	35.94	NA	NA	NA
15	1	default	S5_Bryan Adams	6102485	NA	Captive	MAW	M	A	Winter	Healesville	4.1616438	8.6	NA	NA	NA		NA	NA	CACTAGCT+TCGGATTC	5,317,767	1.00	97.33	2.67	1,606	100	93.24	35.83	NA	NA	NA
16	1	default	S5_Allen	6102488	NA	DFT1	MAU	M	A	Summer	Channel	1.9095890	6.7	NA	1	TRUE	deep	TRUE	55959.750	TTGCAGAC+ATCGTGGT	5,288,435	1.00	96.85	3.15	1,597	100	93.42	35.86	NA	NA	NA
17	1	default	S5_Cyan	6102489	NA	DFT1	MAW	M	A	Winter	Channel	3.3945205	9.1	NA	5	TRUE	deep	TRUE	206006.422	GTAGGAGT+GTCCTAAG	4,138,600	1.00	98.03	1.97	1,250	100	93.67	35.90	NA	NA	NA
18	1	default	S6_Shelob	6102492	NA	DFT2	FAP	F	A	Spring	Channel	4.6219178	7.2	NA	NA	NA		NA	7455.435	GTCATCGA+GATCCACT	3,184,798	1.00	98.21	1.79	962	100	93.90	35.94	NA	NA	NA
19	1	default	S6_Stuttgart	6102493	NA	DFT1	MSP	M	S	Spring	Channel	0.5780822	12.7	NA	1	FALSE		TRUE	3809.806	AGCGTGTT+CTTCGCAA	3,056,475	1.00	97.50	2.50	923	100	93.95	35.95	NA	NA	NA
20	1	default	S7_Fredo	6102496	NA	Captive	MAA	M	A	Autumn	Healesville	2.9945205	8.8	NA	NA	NA		NA	NA	AGTGGATC+TCGATGAC	2,690,632	1.00	97.77	2.23	813	100	93.77	35.93	NA	NA	NA
21	1	default	S7_Piglet	6102497	NA	Captive	FSU	F	S	Summer	Healesville	1.5890411	4.76	NA	NA	NA		NA	NA	ATCACACG+GTAGCGTA	12,134,750	2.00	96.94	3.06	3,665	100	93.34	35.84	NA	NA	NA
22	1	default	S7_Violet	6102498	NA	DFT2	FSU	F	S	Summer	Channel	0.8739726	4.6	NA	1	TRUE	superficial	TRUE	5517.330	CTTGTCGA+CGATGTTC	2,768,682	1.00	98.19	1.81	836	100	93.85	35.93	NA	NA	NA
23	1	default	S7_Willie	6102500	NA	DFT2	MAA	M	A	Autumn	Channel	2.1150685	8.2	NA	2	TRUE	deep	TRUE	1527.480	TGGACTCT+TCTTACGG	4,539,036	1.00	98.30	1.70	1,371	100	93.84	35.93	NA	NA	NA
24	1	default	S8_Orange	6102502	NA	DFT1	FAU	F	A	Summer	Channel	1.9150685	7.1	NA	7	TRUE	superficial	TRUE	43690.461	GTTGACCT+CAGTGCTT	5,072,309	1.00	97.92	2.08	1,532	100	93.36	35.84	NA	NA	NA
25	1	default	S8_ST67	6102505	NA	DFT2	MAW	M	A	Winter	Channel	4.2794521	10.2	NA	1	TRUE	deep	TRUE	102246.432	TTGGACGT+GAAGTGCT	5,476,735	1.00	95.44	4.56	1,654	100	93.21	35.82	NA	NA	NA
26	1	default	S8_Yellow	6102507	NA	DFT2	MAU	M	A	Summer	Channel	1.8657534	10.5	NA	2	TRUE	deep	TRUE	139389.671	GAAGTTGG+CGAACAAC	6,469,324	1.00	97.99	2.01	1,954	100	94.00	35.97	NA	NA	NA
27	1	default	S9_Jane	6102508	NA	DFT1	FAA	F	A	Autumn	Channel	3.1972603	6.1	NA	NA	NA		NA	64646.036	GTTCTCGT+TGTTCCGT	12,489,105	3.00	97.92	2.08	3,772	100	93.54	35.88	NA	NA	NA
28	1	default	S9_Clytemnestra	6102511	NA	DFT1	FSP	F	S	Spring	Channel	1.6602740	6	NA	NA	NA		NA	34738.080	CATGGCTA+CACACATC	6,141,790	1.00	98.00	2.00	1,855	100	93.42	35.85	NA	NA	NA
30	1	default	S10_McMahon	6102514	NA	Captive	MSP	M	S	Spring	Healesville	1.6493151	8.06	NA	NA	NA		NA	NA	AACGTCTG+AATGACGC	3,809,607	1.00	96.73	3.27	1,151	100	93.07	35.79	NA	NA	NA
31	1	default	S10_Logan	6102515	NA	Captive	MAU	M	A	Summer	Healesville	5.6000000	9.02	NA	NA	NA		NA	NA	GTGCCATA+CTCCTAGT	9,606,720	2.00	97.52	2.48	2,901	100	93.25	35.83	NA	NA	NA
32	1	default	S10_Laurasia	6102516	NA	DFT2	MSP	M	S	Spring	Channel	1.6383562	8	NA	NA	NA		NA	2697.152	TACGCTAC+CGTGTGAT	9,716,148	2.00	97.81	2.19	2,934	100	93.97	35.95	NA	NA	NA
33	1	default	S10_Enchilada	6102517	NA	DFT2	MAP	M	A	Spring	Channel	3.6438356	9.3	NA	14	FALSE		FALSE	301301.905	GATTGCTC+CACTTCAC	11,967,844	2.00	97.82	2.18	3,614	100	93.88	35.94	NA	NA	NA
34	1	default	S11_Murdoch/Washington	6102519	NA	DFT2	MAA	M	A	Autumn	Channel	1.4438356	8.9	NA	NA	NA		NA	127176.305	ACTGAGGT+GTCAACAG	2,003,509	0.00	97.82	2.18	605	100	93.39	35.85	NA	NA	NA
35	1	default	S11_Xuki	6102520	NA	Captive	MAA	M	A	Autumn	Monarto	2.0767123	7.95	NA	NA	NA		NA	NA	ATCGCCAT+TGCGTAAC	6,018,884	1.00	97.88	2.12	1,818	100	93.58	35.89	NA	NA	NA
36	1	default	S11_Kalamata	6102521	NA	Captive	MSW	M	S	Winter	Healesville	1.2849315	5.58	NA	NA	NA		NA	NA	GAACATCG+TCGACAAG	6,861,544	1.00	97.76	2.24	2,072	100	93.49	35.87	NA	NA	NA
37	1	default	S12_Abe	6102524	NA	DFT1	MAA	M	A	Autumn	Channel	3.1150685	7.9	NA	1	TRUE	deep	TRUE	13125.000	GTTGTTCG+CCAACTTC	6,446,467	1.00	97.60	2.40	1,947	100	93.65	35.90	NA	NA	NA
38	1	default	S12_Lilac	6102526	NA	DFT1	MAW	M	A	Winter	Channel	3.4109589	9.6	NA	5	TRUE	superficial	TRUE	37293.844	TATCGGTC+TACCGGAT	4,490,241	1.00	98.15	1.85	1,356	100	93.42	35.85	NA	NA	NA
39	1	default	S12_Wally	6102528	NA	Captive	MAA	M	A	Autumn	Monarto	2.1616438	N/A	NA	NA	NA		NA	NA	ACGACTTG+AACAGCGA	3,000,501	1.00	97.96	2.04	906	100	93.70	35.91	NA	NA	NA
40	1	default	S12_Naz	6102529	NA	Captive	MAW	M	A	Winter	Monarto	4.4547945	N/A	NA	NA	NA		NA	NA	CCTTGTAG+CCTTGGAA	5,927,942	1.00	97.40	2.60	1,790	100	93.86	35.94	NA	NA	NA
41	1	default	S13_Humbug	6102530	NA	Captive	MAU	M	A	Summer	Monarto	2.5397260	7.94	NA	NA	NA		NA	NA	GAGCAGTA+TGAGCTGT	1,702,608	0.00	98.07	1.93	514	100	93.58	35.89	NA	NA	NA
42	1	default	S13_Green	6102531	NA	DFT1	MAA	M	A	Autumn	Channel	3.1534247	7.5	NA	2	TRUE	superficial	TRUE	7407.222	GTCGAAGA+CCACATTG	2,623,695	1.00	97.38	2.62	792	100	92.94	35.76	NA	NA	NA
43	1	default	S13_Bump	6102532	NA	DFT1	MAU	M	A	Summer	Channel	3.9068493	9.3	NA	NA	NA		NA	629522.362	AGTTCGTC+ACTCGATC	9,435,053	2.00	98.16	1.84	2,849	100	93.64	35.89	NA	NA	NA
44	1	default	S13_Xerxes	6102535	NA	Captive	MAA	M	A	Autumn	Monarto	2.0767123	N/A	NA	NA	NA		NA	NA	GATAGGCT+CAGTCACA	7,628,257	2.00	98.34	1.66	2,304	100	93.62	35.89	NA	NA	NA
45	1	default	S14_Purple	6102536	NA	DFT1	FAW	F	A	Winter	Channel	2.3945205	6.8	NA	1	TRUE	deep	TRUE	24938.123	CGCTCTAT+TTGCAACG	5,432,182	1.00	97.81	2.19	1,641	100	93.93	35.95	NA	NA	NA
46	1	default	S14_Catani	6102538	NA	Captive	FAU	F	A	Summer	Healesville	2.8630137	6.46	NA	NA	NA		NA	NA	TGAAGACG+ACATGCCA	13,381,058	3.00	96.98	3.02	4,041	100	93.36	35.84	NA	NA	NA
47	2	default	S15_Amelia	6102541	NA	Captive	FAU	F	A	Summer	Healesville	4.7260274	5.85	NA	NA	NA		NA	NA	AACGTCTG+AATGACGC	8,041,818	1.73	96.32	3.68	2,429	100	92.18	35.63	NA	NA	NA
48	2	default	S15_London	6102542	NA	Captive	FAU	F	A	Summer	Healesville	3.7863014	4.8	NA	NA	NA		NA	NA	GTCATCGA+GATCCACT	5,758,111	1.24	96.72	3.28	1,739	100	91.87	35.55	NA	NA	NA
49	2	default	S15_Elyse	6102543	NA	Captive	FAP	F	A	Spring	Healesville	4.5232877	7.88	NA	NA	NA		NA	NA	AGCGTGTT+CTTCGCAA	5,774,637	1.24	96.76	3.24	1,744	100	91.56	35.50	NA	NA	NA
51	2	default	S16_Milla	6102548	NA	Captive	FAU	F	A	Summer	Healesville	2.7671233	5.82	NA	NA	NA		NA	NA	TTGCGAAG+AACACGCT	6,376,997	1.37	97.60	2.40	1,926	100	92.44	35.67	NA	NA	NA
52	2	default	S16_Maleficent	6102549	NA	Captive	FAU	F	A	Summer	Monarto	4.5589041	8.59	NA	NA	NA		NA	NA	TGCTTCCA+CGATCGAT	9,181,668	1.98	98.08	1.92	2,773	100	93.04	35.79	NA	NA	NA
53	2	default	S17_Anirak	6102553	NA	Captive	FSW	F	S	Winter	Healesville	1.3424658	5.42	NA	NA	NA		NA	NA	TGAGCTAG+GAACGGTT	3,740,878	0.81	97.15	2.85	1,130	100	93.03	35.79	NA	NA	NA
54	2	default	S17_Poly	6102554	NA	Captive	FSU	F	S	Summer	Monarto	1.7917808	6.67	NA	NA	NA		NA	NA	CTTAGGAC+ACTCCTAC	11,133,566	2.40	97.85	2.15	3,362	100	92.47	35.68	NA	NA	NA
55	2	default	S17_Methyl	6102556	NA	Captive	FSU	F	S	Summer	Healesville	0.7534247	2.72	NA	NA	NA		NA	NA	TGGTACAG+TCGAGAGT	7,887,375	1.70	97.57	2.43	2,382	100	93.15	35.81	NA	NA	NA
56	2	default	S18_Zara	6102558	NA	Captive	FSW	F	S	Winter	Monarto	1.5150685	4.6	NA	NA	NA		NA	NA	ACCACGAT+GTCTGCAA	7,308,907	1.58	97.70	2.30	2,207	100	92.84	35.75	NA	NA	NA
57	2	default	S18_Maze	6102559	NA	Captive	FSW	F	S	Winter	Healesville	1.4000000	5.82	NA	NA	NA		NA	NA	ACTGAGGT+GTCAACAG	2,984,930	0.64	97.59	2.41	901	100	92.49	35.69	NA	NA	NA
59	2	default	S19_Morocco	6102564	NA	Captive	MAU	M	A	Summer	Monarto	4.5917808	9.82	NA	NA	NA		NA	NA	TCAGGCTT+ATCATGCG	4,879,629	1.05	97.38	2.62	1,474	100	92.63	35.71	NA	NA	NA
60	2	default	S19_Pecorino	6102565	NA	Captive	FSW	F	S	Winter	Healesville	1.1123288	6.02	NA	NA	NA		NA	NA	ATGCCTGT+AGATTGCG	4,857,395	1.05	97.58	2.42	1,467	100	92.64	35.72	NA	NA	NA
62	2	default	S20_Muasy	6102569	NA	Captive	MSU	M	S	Summer	Healesville	1.4219178	6.62	NA	NA	NA		NA	NA	TTCCAAGG+CTACAAGG	10,669,237	2.30	97.64	2.36	3,222	100	92.79	35.74	NA	NA	NA
63	2	default	S20_Fuse	6102570	NA	Captive	MSU	M	S	Summer	Healesville	1.7315068	6.26	NA	NA	NA		NA	NA	GTAGGAGT+GTCCTAAG	10,603,179	2.29	97.34	2.66	3,202	100	92.01	35.58	NA	NA	NA
64	2	default	S20_Joseph	6102571	NA	Captive	MSU	M	S	Summer	Healesville	1.7479452	7.42	NA	NA	NA		NA	NA	CTTCGACT+GATCTTGC	5,642,025	1.22	96.37	3.63	1,704	100	91.87	35.56	NA	NA	NA
65	2	default	S21_Moama	6102576	NA	Captive	MAW	M	A	Winter	Monarto	1.6986301	8.75	NA	NA	NA		NA	NA	TTGGACGT+GAAGTGCT	10,821,895	2.33	96.02	3.98	3,268	100	92.72	35.73	NA	NA	NA
66	2	default	S21_Nauset	6102577	NA	Captive	MSW	M	S	Winter	Healesville	1.3041096	4.036	NA	NA	NA		NA	NA	TATCGGTC+TACCGGAT	8,693,920	1.87	98.13	1.87	2,626	100	92.32	35.65	NA	NA	NA
67	2	default	S21_Graham	6102578	NA	Captive	MSW	M	S	Winter	Healesville	1.3616438	6.4	NA	NA	NA		NA	NA	TACGCCTT+GCTACTCT	3,422,010	0.74	97.72	2.28	1,033	100	92.94	35.76	NA	NA	NA
68	2	default	S22_Yanni	6102581	NA	Captive	MSW	M	S	Winter	Monarto	1.5178082	8.15	NA	NA	NA		NA	NA	CGTGTGTA+CCAACGAA	4,811,453	1.04	97.67	2.33	1,453	100	92.30	35.65	NA	NA	NA
69	2	default	S22_Gherkin	6102582	NA	Captive	MSW	M	S	Winter	Healesville	1.2630137	6.32	NA	NA	NA		NA	NA	CCTTGTAG+CCTTGGAA	5,913,586	1.27	97.05	2.95	1,786	100	92.92	35.77	NA	NA	NA
70	2	default	S22_Stilton	6102584	NA	Captive	MSW	M	S	Winter	Healesville	1.2986301	5.76	NA	NA	NA		NA	NA	CGCAATCT+ACAGGCAT	8,306,537	1.79	96.94	3.06	2,509	100	91.86	35.55	NA	NA	NA
71	2	default	S23_Osiris	6102588	NA	Captive	MSW	M	S	Winter	Healesville	1.1753425	5.34	NA	NA	NA		NA	NA	AGTGGATC+TCGATGAC	3,528,471	0.76	97.24	2.76	1,066	100	93.08	35.79	NA	NA	NA
73	2	default	S23_Quince	6102590	NA	Captive	MSW	M	S	Winter	Healesville	1.3835616	N/A	NA	NA	NA		NA	NA	AACCGTTC+CTAGCTCA	7,247,793	1.56	97.63	2.37	2,189	100	92.28	35.63	NA	NA	NA
74	2	default	S24_Malfoy	6102595	NA	Captive	MAW	M	A	Winter	Healesville	5.1041096	8.02	NA	NA	NA		NA	NA	TAACCGGT+ATCGTCTC	6,908,416	1.49	97.09	2.91	2,086	100	91.90	35.56	NA	NA	NA
75	2	default	S24_Licorice	6102596	NA	Captive	FSA	F	S	Autumn	Healesville	1.8246575	5.94	NA	NA	NA		NA	NA	TTGCAGAC+ATCGTGGT	4,475,461	0.96	95.05	4.95	1,352	100	91.42	35.46	NA	NA	NA
76	2	default	S25_Tofu	6102599	NA	Captive	FAU	F	A	Summer	Healesville	2.7452055	N/A	NA	NA	NA		NA	NA	GAGACGAT+ACCGGTTA	12,024,421	2.59	95.37	4.63	3,631	100	93.06	35.79	NA	NA	NA
77	2	default	S25_Colette	6102600	NA	Captive	FAP	F	A	Spring	Healesville	6.6273973	7.52	NA	NA	NA		NA	NA	CTTGTCGA+CGATGTTC	10,751,797	2.32	97.49	2.51	3,247	100	92.77	35.74	NA	NA	NA
78	2	default	S25_CannonBall	6102601	NA	Captive	MSW	M	S	Winter	Healesville	1.1479452	6.54	NA	NA	NA		NA	NA	TACGCTAC+CGTGTGAT	13,015,081	2.81	97.36	2.64	3,931	100	92.61	35.70	NA	NA	NA
79	2	default	S25_Cornelius	6102602	NA	Captive	MAU	M	A	Summer	Healesville	3.6904110	7.74	NA	NA	NA		NA	NA	GAATCCGA+AGCTAGTG	3,336,983	0.72	97.10	2.90	1,008	100	93.09	35.79	NA	NA	NA
80	2	default	S25_Mardook	6102603	NA	Captive	MAW	M	A	Winter	Healesville	4.3013699	9.16	NA	NA	NA		NA	NA	GCACAACT+CTGAACGT	6,826,313	1.47	97.66	2.34	2,062	100	92.67	35.71	NA	NA	NA
81	2	default	S25_Negroni	6102604	NA	Captive	FSP	F	S	Spring	Healesville	1.6767123	5.86	NA	NA	NA		NA	NA	TGTGACTG+AGCCTATC	4,566,180	0.98	97.68	2.32	1,379	100	92.13	35.61	NA	NA	NA

col <- as.character(as.numeric(grepl("DFT",ssCvDAV$Category)))
col <- gsub("0","lightblue",col)
col <- gsub("1","pink",col)
mds<-plotMDS(x1,pch=19,cex=3,col=col,main="MDS plot: Captive vs DFTD infected devils")
mtext("Captive=Blue, DFTD Infected=Red")

ssCvDAV$DFTDStat <- grepl("DFT",ssCvDAV$Category)
ssCvDAV$CapLoc <- grepl("Channel",ssCvDAV$Location)
ssCvDAV$MvF <- factor(ssCvDAV$Sex)
ssCvDAV$SubAdvAd <- as.numeric(factor(ssCvDAV$Age,levels=c("S","A"),ordered = TRUE))
ssCvDAV$Season <- factor(ssCvDAV$Season)
str(ssCvDAV)

## 'data.frame':    76 obs. of  35 variables:
##  $ Lane                   : int  1 1 1 1 1 1 1 1 1 1 ...
##  $ Project                : chr  "default" "default" "default" "default" ...
##  $ Client.ID              : chr  "S1_Sarafina" "S1_Lucy" "S2_B99008" "S2_Bonney" ...
##  $ Sample                 : int  6102463 6102464 6102468 6102469 6102471 6102472 6102473 6102474 6102476 6102477 ...
##  $ X                      : logi  NA NA NA NA NA NA ...
##  $ Category               : chr  "DFT1" "DFT1" "Captive" "Captive" ...
##  $ Sex.Age.Season         : chr  "FAP" "FAU" "FAU" "FAU" ...
##  $ Sex                    : chr  "F" "F" "F" "F" ...
##  $ Age                    : chr  "A" "A" "A" "A" ...
##  $ Season                 : Factor w/ 4 levels "Autumn","Spring",..: 2 3 3 3 3 3 1 4 4 1 ...
##  $ Location               : chr  "Channel" "Channel" "Monarto" "Healesville" ...
##  $ Agebyyear              : num  2.62 2.92 3.74 2.59 2.93 ...
##  $ Weight                 : chr  "6.4" "7.1" "8.12" "5.4" ...
##  $ X.1                    : logi  NA NA NA NA NA NA ...
##  $ Number.of.Tumours      : int  NA 3 NA NA 4 1 1 1 NA NA ...
##  $ Ulcerated              : logi  NA TRUE NA NA TRUE TRUE ...
##  $ UlcerationDepth        : chr  "" "deep" "" "" ...
##  $ Secondary.infection    : logi  NA TRUE NA NA TRUE FALSE ...
##  $ Total.tumour.volume    : num  95141 700825 NA NA 103816 ...
##  $ Barcode.sequence       : chr  "ATCCAGAG+TAACGTCG" "AAGCACTG+AGGTCAAC" "GCTATCCT+AACAGGTG" "ACGTTCAG+AGTTGTGC" ...
##  $ PF.Clusters            : chr  "6,491,488" "6,121,785" "5,338,890" "14,968,459" ...
##  $ X..of.the.lane         : num  1 1 1 3 0 1 1 1 1 3 ...
##  $ X..Perfect.barcode     : num  74.1 98 98.1 97.8 98 ...
##  $ X..One.mismatch.barcode: num  25.93 2 1.92 2.18 2.04 ...
##  $ Yield..Mbases.         : chr  "1,960" "1,849" "1,612" "4,520" ...
##  $ X..PF.Cluster          : int  100 100 100 100 100 100 100 100 100 100 ...
##  $ X.....Q30              : num  93.7 93.8 93.7 93.7 93.6 ...
##  $ Mean.Quality           : num  35.9 35.9 35.9 35.9 35.9 ...
##  $ X.2                    : logi  NA NA NA NA NA NA ...
##  $ X.3                    : logi  NA NA NA NA NA NA ...
##  $ X.4                    : logi  NA NA NA NA NA NA ...
##  $ DFTDStat               : logi  TRUE TRUE FALSE FALSE TRUE TRUE ...
##  $ CapLoc                 : logi  TRUE TRUE FALSE FALSE TRUE TRUE ...
##  $ MvF                    : Factor w/ 2 levels "F","M": 1 1 1 1 1 1 2 1 1 2 ...
##  $ SubAdvAd               : num  2 2 2 2 2 2 2 2 2 2 ...

dds <- DESeqDataSetFromMatrix(countData = x1 , colData = ssCvDAV, design = ~ Season + SubAdvAd + MvF + DFTDStat)

## converting counts to integer mode

##   the design formula contains one or more numeric variables with integer values,
##   specifying a model with increasing fold change for higher values.
##   did you mean for this to be a factor? if so, first convert
##   this variable to a factor using the factor() function

res <- DESeq(dds)

## estimating size factors

## estimating dispersions

## gene-wise dispersion estimates

## mean-dispersion relationship

## final dispersion estimates

## fitting model and testing

## -- replacing outliers and refitting for 1 genes
## -- DESeq argument 'minReplicatesForReplace' = 7 
## -- original counts are preserved in counts(dds)

## estimating dispersions

## fitting model and testing

z<- results(res)
vsd <- vst(dds, blind=FALSE,nsub=nrow(x1)-50)
zz<-cbind(as.data.frame(z),assay(vsd))
dge<-as.data.frame(zz[order(zz$pvalue),])
dgeCvDAV <- dge
sig <- subset(dge,padj<0.05)
sig1_up <- rownames(subset(sig,log2FoldChange>0))
sig1_dn <- rownames(subset(sig,log2FoldChange<0))
length(sig1_up)

## [1] 95

length(sig1_dn)

## [1] 77

maplot(dgeCvDAV,"Captive vs DFTD infected devils")

make_volcano(dgeCvDAV,"Captive vs DFTD infected devils")

sig[1:200,1:6] %>%
  kbl(caption="Comparison of Captive vs DFTD infected devils") %>%
  kable_paper("hover", full_width = F)

Comparison of Captive vs DFTD infected devils

	baseMean	log2FoldChange	lfcSE	stat	pvalue	padj
CCR2	6595.31407	1.7628439	0.1894639	9.304381	0.0000000	0.0000000
CCR7	20785.67540	-1.6900738	0.1824349	-9.263985	0.0000000	0.0000000
Spi1	48601.23340	1.7169624	0.1888967	9.089424	0.0000000	0.0000000
EPX	4106.91806	3.1410435	0.3649473	8.606840	0.0000000	0.0000000
CD68	6236.28055	1.5313916	0.1839810	8.323639	0.0000000	0.0000000
CCR6	1142.29906	-1.7638245	0.2119380	-8.322361	0.0000000	0.0000000
CD1d	1670.11674	-1.7395569	0.2123495	-8.191950	0.0000000	0.0000000
IL10RA	13131.82917	-1.5401583	0.1927183	-7.991760	0.0000000	0.0000000
CD177	7563.42898	4.2907346	0.5495403	7.807861	0.0000000	0.0000000
SIRPA	310.79955	2.4243983	0.3142203	7.715600	0.0000000	0.0000000
IL7RA	7625.62756	-1.5767051	0.2051315	-7.686314	0.0000000	0.0000000
CD14	59164.24661	1.7361596	0.2307464	7.524102	0.0000000	0.0000000
MRC1	1905.57913	1.5799426	0.2103578	7.510738	0.0000000	0.0000000
CD28	1204.61105	-1.9386374	0.2585637	-7.497717	0.0000000	0.0000000
CATH3	70782.75127	3.9138615	0.5267362	7.430401	0.0000000	0.0000000
TGFb1	72798.56452	1.1546551	0.1585888	7.280810	0.0000000	0.0000000
C3	59510.17158	1.5245188	0.2115578	7.206159	0.0000000	0.0000000
ITGAX	7071.19445	0.9302686	0.1301772	7.146173	0.0000000	0.0000000
TNFSF13	669.05251	1.1184076	0.1590279	7.032775	0.0000000	0.0000000
TLR3	610.86609	-1.1555779	0.1656864	-6.974489	0.0000000	0.0000000
IFIH1	2295.17103	-0.8283654	0.1198019	-6.914458	0.0000000	0.0000000
TNFRSF1A	11338.94883	1.0872456	0.1582227	6.871615	0.0000000	0.0000000
Tap1	8878.07862	-1.5175057	0.2266103	-6.696544	0.0000000	0.0000000
TCRm	5713.97921	-1.4101570	0.2106794	-6.693379	0.0000000	0.0000000
FGF2	75.19974	-1.5390625	0.2353199	-6.540299	0.0000000	0.0000000
STAT1	10514.83671	-0.9653122	0.1508351	-6.399786	0.0000000	0.0000000
CEBPA	2170.22294	1.2652016	0.1995792	6.339344	0.0000000	0.0000000
TNFSF8	589.88958	-1.0666587	0.1690706	-6.308956	0.0000000	0.0000000
PTPRC	48346.24334	-1.0048798	0.1592967	-6.308226	0.0000000	0.0000000
CCLD14	318.72037	1.9966513	0.3186352	6.266262	0.0000000	0.0000000
IL18R1	2221.09499	1.0150233	0.1632209	6.218709	0.0000000	0.0000000
IL18	2072.65292	1.8732642	0.3019994	6.202875	0.0000000	0.0000000
STAT4	1723.35298	-1.1763912	0.1915870	-6.140245	0.0000000	0.0000000
CCR5	381.17965	0.9863201	0.1612636	6.116198	0.0000000	0.0000000
IRF4	931.91435	-1.4154399	0.2315111	-6.113918	0.0000000	0.0000000
CCLD13	377.73327	1.6841255	0.2768392	6.083406	0.0000000	0.0000000
LTB	21724.42164	-1.1167049	0.1865801	-5.985124	0.0000000	0.0000000
FCER2	78.30588	1.9493500	0.3300478	5.906265	0.0000000	0.0000000
MPO	487.56014	3.3312762	0.5647298	5.898885	0.0000000	0.0000000
IL2RA	278.50289	-1.1604301	0.1969970	-5.890599	0.0000000	0.0000000
CXCL16	5847.10141	1.4212656	0.2413738	5.888236	0.0000000	0.0000000
KIT	3614.55293	-1.2016988	0.2062461	-5.826530	0.0000000	0.0000000
TXLNA	4753.20519	0.5177159	0.0889270	5.821808	0.0000000	0.0000000
CD3e	18974.73756	-0.9390289	0.1623856	-5.782710	0.0000000	0.0000000
SELL	92179.26381	0.8164635	0.1412998	5.778235	0.0000000	0.0000000
IL5Ra	1587.67232	-1.2694307	0.2218427	-5.722211	0.0000000	0.0000001
CXCR4	23987.24605	-0.6301161	0.1112654	-5.663180	0.0000000	0.0000001
IgM	90902.80325	-1.0635994	0.1893216	-5.617951	0.0000000	0.0000001
IL21R	4772.00660	-1.0316789	0.1843718	-5.595645	0.0000000	0.0000001
CXCR3	1639.34455	0.9731334	0.1740210	5.592046	0.0000000	0.0000001
CD3d	6106.56175	-0.9746055	0.1750196	-5.568550	0.0000000	0.0000001
IRF9	23943.62163	0.5318834	0.0960078	5.540000	0.0000000	0.0000002
IL16	19995.70779	-0.3991385	0.0724832	-5.506638	0.0000000	0.0000002
IRF3	15893.37284	0.7550161	0.1391981	5.424040	0.0000001	0.0000003
CD40LG	995.14219	-0.7062460	0.1314943	-5.370926	0.0000001	0.0000004
CCR4	1635.38696	-1.1050368	0.2057567	-5.370599	0.0000001	0.0000004
CTLA4	291.48526	-1.1646254	0.2175880	-5.352433	0.0000001	0.0000004
EEF1A1	30.42759	-2.2454000	0.4210478	-5.332887	0.0000001	0.0000004
TCRa	9885.66057	-0.7949798	0.1508248	-5.270884	0.0000001	0.0000006
GUSB	5430.86734	0.6561268	0.1256013	5.223887	0.0000002	0.0000008
FLT3	1546.95694	-1.1779143	0.2297419	-5.127120	0.0000003	0.0000013
CCR1	580.58192	1.0664381	0.2081348	5.123787	0.0000003	0.0000013
NLRP3	10200.30811	0.9881662	0.1930455	5.118826	0.0000003	0.0000013
CXCR2	22859.13461	1.1521996	0.2280400	5.052620	0.0000004	0.0000018
IL1RN	2740.26174	1.6201851	0.3255919	4.976122	0.0000006	0.0000026
CD244	2049.01235	-0.9532676	0.1942448	-4.907558	0.0000009	0.0000037
CFI	576.15605	-0.7734236	0.1579854	-4.895538	0.0000010	0.0000039
Saha-UA	326568.98815	1.1620194	0.2377119	4.888351	0.0000010	0.0000039
IL34	1265.08113	1.6305350	0.3339322	4.882832	0.0000010	0.0000040
PRF1	2057.85268	-1.2838806	0.2632662	-4.876740	0.0000011	0.0000041
VEGF-A	335.33370	-0.9981717	0.2068060	-4.826610	0.0000014	0.0000052
Lgals3	23573.84408	0.7661443	0.1599527	4.789817	0.0000017	0.0000061
CD40	3431.56768	-0.8314972	0.1752767	-4.743912	0.0000021	0.0000076
ENPP3	45.82988	3.4732672	0.7330787	4.737919	0.0000022	0.0000077
TNFSF18	41.48744	3.7583786	0.8041913	4.673488	0.0000030	0.0000104
CXCR5	2200.94792	-0.9768638	0.2106687	-4.636968	0.0000035	0.0000123
KLRK1	236.61904	-1.1047489	0.2390111	-4.622166	0.0000038	0.0000130
cGAS	457.72297	0.6876746	0.1520666	4.522193	0.0000061	0.0000207
CCL24	1594.89608	1.0851567	0.2412950	4.497221	0.0000069	0.0000230
Saha-UD	910589.56488	1.3871469	0.3092074	4.486137	0.0000073	0.0000239
CIITA	1959.50372	-0.7996516	0.1793462	-4.458705	0.0000082	0.0000269
CD37	36277.54190	0.5363980	0.1210850	4.429931	0.0000094	0.0000303
PRG3	10936.84771	-0.9726263	0.2197290	-4.426481	0.0000096	0.0000305
TCRd	298.43726	-1.1394236	0.2577267	-4.421054	0.0000098	0.0000309
CXCL10	19.92272	-1.9681184	0.4485562	-4.387674	0.0000115	0.0000356
CSF1	1793.55358	1.1655950	0.2663464	4.376237	0.0000121	0.0000367
GATA2	1779.73980	0.8343123	0.1906553	4.376024	0.0000121	0.0000367
CD96	256.43242	-1.3931349	0.3189889	-4.367346	0.0000126	0.0000377
ICOS	683.63416	-0.9139098	0.2095503	-4.361292	0.0000129	0.0000384
TLR2	7784.66544	0.7902484	0.1824227	4.331963	0.0000148	0.0000429
LTa	311.33897	-0.8096034	0.1868992	-4.331765	0.0000148	0.0000429
Saha-UK	202061.16063	0.9562961	0.2221303	4.305114	0.0000167	0.0000479
IL17D	279.03183	-1.0183420	0.2370458	-4.295971	0.0000174	0.0000494
IgA	20533.60052	-0.8926403	0.2080516	-4.290476	0.0000178	0.0000501
CD4	3926.32061	-0.5744018	0.1345503	-4.269048	0.0000196	0.0000546
IL3Ra	8541.32152	0.8266584	0.1950334	4.238547	0.0000225	0.0000615
IL27Ra	1460.74743	-0.7578981	0.1788534	-4.237537	0.0000226	0.0000615
NLRP1	813.45656	-0.6281317	0.1492535	-4.208490	0.0000257	0.0000693
OAZ1	436243.29529	1.1854687	0.2822569	4.199964	0.0000267	0.0000709
CCR8	233.73296	-1.0146397	0.2416606	-4.198615	0.0000269	0.0000709
MMP9	86718.33560	0.8690983	0.2088778	4.160798	0.0000317	0.0000829
TLR6	3701.06582	0.6400968	0.1569884	4.077350	0.0000456	0.0001179
NLCR5	8935.68079	-0.4778819	0.1174987	-4.067125	0.0000476	0.0001220
MME	83.69032	2.8028358	0.6947003	4.034597	0.0000547	0.0001388
LAMP1	27646.70111	0.4968364	0.1268988	3.915216	0.0000903	0.0002271
Saha-UC	445789.17402	0.9721053	0.2491490	3.901703	0.0000955	0.0002379
IL12Rb1	1149.75656	-0.7842206	0.2020325	-3.881655	0.0001037	0.0002560
MS4A1	9550.15427	-0.7761226	0.2058008	-3.771232	0.0001624	0.0003971
CD8a	16139.16908	-0.8342460	0.2216477	-3.763838	0.0001673	0.0004053
NOD2	2114.61644	0.9388651	0.2543915	3.690631	0.0002237	0.0005369
TCF3	28653.36118	0.7816314	0.2130729	3.668375	0.0002441	0.0005806
GATA1	240511.67585	1.8590753	0.5121881	3.629673	0.0002838	0.0006689
CCLD4	12.03493	2.8798841	0.8002184	3.598873	0.0003196	0.0007467
IgG	27016.42383	-0.6329984	0.1769999	-3.576265	0.0003485	0.0008071
CXCR1	21268.15185	0.7931668	0.2222147	3.569371	0.0003578	0.0008190
TNFSF11	109.63951	-1.1007786	0.3085949	-3.567067	0.0003610	0.0008190
CXCL12	62.46433	2.3727275	0.6654415	3.565644	0.0003630	0.0008190
IL13Ra1	5086.15677	0.7323276	0.2066633	3.543579	0.0003947	0.0008831
FCGRT	23963.28246	0.5179049	0.1470108	3.522903	0.0004268	0.0009470
NFKB1	8351.22345	-0.2921498	0.0832019	-3.511335	0.0004459	0.0009809
IL13	52.61872	1.3313022	0.3828160	3.477656	0.0005058	0.0011036
CD22	14182.64661	-0.5313198	0.1539821	-3.450531	0.0005595	0.0012107
IL1A	66.27029	1.6590112	0.4970575	3.337665	0.0008449	0.0018133
CD19	4205.36702	-0.7153143	0.2161675	-3.309074	0.0009361	0.0019929
EIF2AK2	563.99610	-0.5462142	0.1653914	-3.302556	0.0009581	0.0020235
CCR3	10070.10610	-0.6353389	0.1932225	-3.288121	0.0010086	0.0021132
EGFR	31.56224	2.8048318	0.8582738	3.267992	0.0010831	0.0022516
HPRT	2297.60196	0.4026545	0.1241855	3.242364	0.0011854	0.0024449
IL22	29.73360	2.6576227	0.8209645	3.237196	0.0012071	0.0024704
B2M	255730.15462	0.7834729	0.2424410	3.231602	0.0012310	0.0024998
CCL25	22.38266	2.1864533	0.6780864	3.224447	0.0012622	0.0025436
IL9	13.55122	2.9848147	0.9401041	3.174983	0.0014985	0.0029969
CCLD1	34.13033	1.7488189	0.5518121	3.169229	0.0015284	0.0030339
IL1B	11185.11259	0.7673939	0.2423074	3.167026	0.0015401	0.0030342
PDGFA	1679.68658	0.9585078	0.3048941	3.143740	0.0016680	0.0032619
LBR	11673.65753	-0.5103961	0.1637741	-3.116465	0.0018303	0.0035530
CXCLD1	27.12006	2.2073198	0.7104699	3.106845	0.0018910	0.0036439
CCL20	31.37279	2.4030741	0.7811868	3.076184	0.0020967	0.0040111
FCMR	42443.99542	-0.7353009	0.2435269	-3.019383	0.0025329	0.0048107
FCGR3	3391.15896	0.8132064	0.2750600	2.956469	0.0031118	0.0058399
CD163	1059.81784	0.6119408	0.2071436	2.954186	0.0031349	0.0058399
IL17C	23.95429	1.2331731	0.4175185	2.953577	0.0031411	0.0058399
RPS29	25020.74854	0.6529513	0.2239404	2.915737	0.0035485	0.0065511
TCRb	15926.45146	-0.5459267	0.1884074	-2.897586	0.0037605	0.0068624
CXCL11	13.56342	3.3359497	1.1515719	2.896866	0.0037691	0.0068624
IL15	62.43146	1.1388644	0.3976160	2.864232	0.0041802	0.0075588
CCN2	66.79228	1.1552848	0.4103421	2.815419	0.0048714	0.0087486
IRF5	5351.44309	0.4302444	0.1541647	2.790811	0.0052576	0.0093785
CCLD3	54.40737	1.4559248	0.5221450	2.788353	0.0052977	0.0093865
CD83	127.33281	-0.7455680	0.2696338	-2.765114	0.0056903	0.0100149
FASLG	426.10547	-0.6578068	0.2386281	-2.756619	0.0058402	0.0101565
C5	107.20039	1.0367952	0.3761682	2.756201	0.0058477	0.0101565
IL1RL1	776.13709	-0.5546638	0.2021642	-2.743630	0.0060764	0.0104848
RPLP0	156083.89647	0.4839232	0.1775205	2.726013	0.0064104	0.0109893
C4A	336.08605	0.5017211	0.1845782	2.718204	0.0065637	0.0111795
Foxp3	252.26349	-0.6708044	0.2498293	-2.685051	0.0072519	0.0122447
STAT5A	18586.09438	0.2651481	0.0988005	2.683671	0.0072819	0.0122447
Saha-DAA	97439.75912	-0.3218055	0.1214113	-2.650539	0.0080363	0.0134278
STING1	1079.34186	-0.4374169	0.1702508	-2.569250	0.0101919	0.0169224
TNF	114.79779	-0.6100376	0.2432089	-2.508287	0.0121318	0.0200175
TLR4	3048.61911	0.5532104	0.2223160	2.488396	0.0128321	0.0210414
IL4R	8400.10036	-0.3617917	0.1456036	-2.484771	0.0129635	0.0211233
XCR1	57.11254	0.7321103	0.2948943	2.482619	0.0130420	0.0211233
CXCLD2	25.47679	1.7362336	0.7113817	2.440650	0.0146609	0.0236004
IL10	31.95184	0.9174660	0.3795057	2.417529	0.0156263	0.0248753
IL23R	100.05620	0.6950963	0.2875650	2.417180	0.0156413	0.0248753
Rorc	290.94293	-0.4334494	0.1803270	-2.403686	0.0162307	0.0256581
STAT6	18756.91043	0.1793280	0.0753319	2.380506	0.0172889	0.0271682
IFNLR1	95.27446	-0.8812767	0.3820964	-2.306425	0.0210869	0.0329405
IL11Ra	1481.26646	-0.3499738	0.1528059	-2.290316	0.0220030	0.0341693
CD59	18742.52151	0.6715034	0.2935864	2.287243	0.0221816	0.0342453
TNFRSF1B	7406.74606	0.2633392	0.1215656	2.166231	0.0302935	0.0464971
NA	NA	NA	NA	NA	NA	NA
NA.1	NA	NA	NA	NA	NA	NA
NA.2	NA	NA	NA	NA	NA	NA
NA.3	NA	NA	NA	NA	NA	NA
NA.4	NA	NA	NA	NA	NA	NA
NA.5	NA	NA	NA	NA	NA	NA
NA.6	NA	NA	NA	NA	NA	NA
NA.7	NA	NA	NA	NA	NA	NA
NA.8	NA	NA	NA	NA	NA	NA
NA.9	NA	NA	NA	NA	NA	NA
NA.10	NA	NA	NA	NA	NA	NA
NA.11	NA	NA	NA	NA	NA	NA
NA.12	NA	NA	NA	NA	NA	NA
NA.13	NA	NA	NA	NA	NA	NA
NA.14	NA	NA	NA	NA	NA	NA
NA.15	NA	NA	NA	NA	NA	NA
NA.16	NA	NA	NA	NA	NA	NA
NA.17	NA	NA	NA	NA	NA	NA
NA.18	NA	NA	NA	NA	NA	NA
NA.19	NA	NA	NA	NA	NA	NA
NA.20	NA	NA	NA	NA	NA	NA
NA.21	NA	NA	NA	NA	NA	NA
NA.22	NA	NA	NA	NA	NA	NA
NA.23	NA	NA	NA	NA	NA	NA
NA.24	NA	NA	NA	NA	NA	NA
NA.25	NA	NA	NA	NA	NA	NA
NA.26	NA	NA	NA	NA	NA	NA
NA.27	NA	NA	NA	NA	NA	NA

write.table(dge,file="dgeCvDAV.tsv",sep="\t")

mx <- sig[,7:ncol(sig)]
mx <- head(mx,30)
colfunc <- colorRampPalette(c("blue", "white", "red"))
heatmap.2(as.matrix(mx),trace="none",scale="row",
          col=colfunc(25),ColSideColors=col,mar=c(5,12)) 

ssHS <- filter(ss, Category %in% c('Healthy', 'Healthy/Wild'))

ssHSA <- ssHS[which(grepl("S",ssHS$Age) + grepl("A",ssHS$Age)>0),]
ssHSA

##    Lane Project     Client.ID  Sample  X     Category Sex.Age.Season Sex Age
## 1     1 default S1_LauraGrace 6102461 NA Healthy/Wild            FAP   F   A
## 2     1 default   S1_Turmeric 6102465 NA Healthy/Wild            FAU   F   A
## 3     1 default  S1_Singapore 6102466 NA      Healthy            FAA   F   A
## 4     1 default    S2_Belinda 6102467 NA Healthy/Wild            FAU   F   A
## 5     1 default      S2_Rosie 6102470 NA Healthy/Wild            FAU   F   A
## 6     1 default       S3_ST62 6102475 NA Healthy/Wild            FAW   F   A
## 7     1 default    S4_Kerchak 6102478 NA      Healthy            MAA   M   A
## 8     1 default    S4_Duffman 6102479 NA      Healthy            MAA   M   A
## 9     1 default      S4_Hogan 6102481 NA      Healthy            MAA   M   A
## 10    1 default      S5_Silky 6102486 NA Healthy/Wild            MAW   M   A
## 11    1 default    S5_Longley 6102487 NA Healthy/Wild            MAU   M   A
## 12    1 default    S6_Myanmar 6102490 NA      Healthy            FAP   F   A
## 13    1 default     S6_Pakana 6102491 NA      Healthy            MSP   M   S
## 14    1 default     S6_Hobbit 6102495 NA      Healthy            MSP   M   S
## 15    1 default      S7_Lieke 6102499 NA Healthy/Wild            FSU   F   S
## 16    1 default   S7_Smithers 6102501 NA Healthy/Wild            MAA   M   A
## 17    1 default     S8_Vinnie 6102503 NA Healthy/Wild            MAU   M   A
## 18    1 default  S8_Struggles 6102504 NA Healthy/Wild            MAW   M   A
## 19    1 default     S8_SusieQ 6102506 NA Healthy/Wild            FAU   F   A
## 20    1 default     S9_Majula 6102509 NA      Healthy            FAA   F   A
## 21    1 default     S9_Snoopy 6102510 NA Healthy/Wild            MSW   M   S
## 22    1 default     S9_Sarabi 6102513 NA Healthy/Wild            FSP   F   S
## 23    1 default     S11_Baloo 6102518 NA Healthy/Wild            MAA   M   A
## 24    1 default S11_DanRhodes 6102523 NA Healthy/Wild            MSW   M   S
## 25    1 default     S12_Robin 6102525 NA Healthy/Wild            MAA   M   A
## 26    1 default    S12_Eeyore 6102527 NA Healthy/Wild            MAW   M   A
## 27    1 default    S13_Pangea 6102533 NA Healthy/Wild            MAU   M   A
## 28    1 default    S13_Finbar 6102534 NA Healthy/Wild            MAA   M   A
## 29    2 default    S22_Ginger 6102586 NA Healthy/Wild            MSW   M   S
## 30    2 default       S24_Boy 6102593 NA Healthy/Wild            MAA   M   A
## 31    2 default    S24_Shazza 6102598 NA      Healthy            FAW   F   A
##    Season Location Agebyyear Weight X.1 Number.of.Tumours Ulcerated
## 1  Spring  Channel 2.6219178    7.2  NA                NA        NA
## 2  Summer  Channel 1.9095890    7.8  NA                NA        NA
## 3  Autumn  Channel 5.0000000    8.6  NA                NA        NA
## 4  Summer  Channel 1.8657534    6.6  NA                NA        NA
## 5  Summer  Channel 1.8904110    7.4  NA                NA        NA
## 6  Winter  Channel 3.2712329   10.2  NA                NA        NA
## 7  Autumn  Channel 4.1863014   11.6  NA                NA        NA
## 8  Autumn  Channel 3.1123288   12.4  NA                NA        NA
## 9  Autumn  Channel 3.1616438   11.6  NA                NA        NA
## 10 Winter  Channel 2.2849315    9.5  NA                NA        NA
## 11 Summer  Channel 2.9315068    9.7  NA                NA        NA
## 12 Spring  Channel 5.5726027    7.6  NA                NA        NA
## 13 Spring  Channel 1.6712329    7.8  NA                NA        NA
## 14 Spring  Channel 1.6273973      9  NA                NA        NA
## 15 Summer  Channel 0.8657534    4.9  NA                NA        NA
## 16 Autumn  Channel 3.1013699    8.9  NA                NA        NA
## 17 Summer  Channel 2.9150685    8.8  NA                NA        NA
## 18 Winter  Channel 3.2821918   11.1  NA                NA        NA
## 19 Summer  Channel 1.8712329    7.1  NA                NA        NA
## 20 Autumn  Channel 3.1150685      6  NA                NA        NA
## 21 Winter  Channel 1.2794521    6.3  NA                NA        NA
## 22 Spring  Channel 1.6328767    5.8  NA                NA        NA
## 23 Autumn  Channel 3.1726027   11.5  NA                NA        NA
## 24 Winter  Channel 1.3917808    6.8  NA                NA        NA
## 25 Autumn  Channel 2.1972603    7.7  NA                NA        NA
## 26 Winter  Channel 3.2794521      9  NA                NA        NA
## 27 Summer  Channel 1.8794521    7.7  NA                NA        NA
## 28 Autumn  Channel 2.1150685    6.8  NA                NA        NA
## 29 Winter  Channel 1.2712329    7.6  NA                NA        NA
## 30 Autumn  Channel 3.1452055    8.3  NA                NA        NA
## 31 Winter  Channel 2.0273973    6.2  NA                NA        NA
##    UlcerationDepth Secondary.infection Total.tumour.volume  Barcode.sequence
## 1                                   NA                  NA AGAGTAGC+AAGGCGTA
## 2                                   NA                  NA GAAGGAAG+GACGTCAT
## 3                                   NA                  NA ACAGCTCA+TACTGCTC
## 4                                   NA                  NA GCAAGATC+AGTCGAAG
## 5                                   NA                  NA CGCATGAT+AAGCCTGA
## 6                                   NA                  NA TGAGCTAG+GAACGGTT
## 7                                   NA                  NA CGGCTAAT+CTCGTTCT
## 8                                   NA                  NA GACGATCT+TCGTGCAT
## 9                                   NA                  NA TACGCCTT+GCTACTCT
## 10                                  NA                  NA CTGTTGAC+ACCTCAGT
## 11                                  NA                  NA CGCAATCT+ACAGGCAT
## 12                                  NA                  NA GCACAACT+CTGAACGT
## 13                                  NA                  NA GTTACGCA+ATGGCGAT
## 14                                  NA                  NA ATATGCGC+ACGATCAG
## 15                                  NA                  NA TTCCAAGG+CTACAAGG
## 16                                  NA                  NA GCGTCATT+CAGACGTT
## 17                                  NA                  NA CGACGTTA+CTCTGGAT
## 18                                  NA                  NA ATCGATCG+TGGAAGCA
## 19                                  NA                  NA CAGTCCAA+CGGTAATC
## 20                                  NA                  NA ACTCTCGA+CTGTACCA
## 21                                  NA                  NA AACTGAGC+CAATCAGG
## 22                                  NA                  NA CTTAGGAC+ACTCCTAC
## 23                                  NA                  NA CTGATCGT+GCGCATAT
## 24                                  NA                  NA TCGCTGTT+CAAGTCGT
## 25                                  NA                  NA ATGACGTC+CTTCCTTC
## 26                                  NA                  NA TCAGGCTT+ATCATGCG
## 27                                  NA                  NA ACCACGAT+GTCTGCAA
## 28                                  NA                  NA GAGACGAT+ACCGGTTA
## 29                                  NA                  NA ACAGCTCA+TACTGCTC
## 30                                  NA                  NA CGACGTTA+CTCTGGAT
## 31                                  NA                  NA GAGCAGTA+TGAGCTGT
##    PF.Clusters X..of.the.lane X..Perfect.barcode X..One.mismatch.barcode
## 1    4,511,702           1.00              98.29                    1.71
## 2    6,479,047           1.00              97.28                    2.72
## 3    4,481,213           1.00              97.99                    2.01
## 4    3,296,547           1.00              97.40                    2.60
## 5    4,556,781           1.00              95.70                    4.30
## 6    4,579,681           1.00              96.67                    3.33
## 7    6,886,511           1.00              97.80                    2.20
## 8    8,301,336           2.00              97.75                    2.25
## 9    2,990,345           1.00              97.77                    2.23
## 10   5,617,260           1.00              97.81                    2.19
## 11   2,937,660           1.00              97.71                    2.29
## 12   9,604,175           2.00              98.14                    1.86
## 13   4,383,521           1.00              98.09                    1.91
## 14   7,342,414           2.00              97.00                    3.00
## 15   1,800,468           0.00              97.79                    2.21
## 16   6,731,177           1.00              98.04                    1.96
## 17   7,901,958           2.00              98.02                    1.98
## 18   5,257,820           1.00              95.13                    4.87
## 19     454,240           0.00              97.95                    2.05
## 20   8,490,270           2.00              97.99                    2.01
## 21   1,709,078           0.00              98.04                    1.96
## 22   3,945,235           1.00              98.11                    1.89
## 23  10,923,019           2.00              98.07                    1.93
## 24   5,076,673           1.00              97.41                    2.59
## 25   6,756,342           1.00              97.12                    2.88
## 26   3,071,283           1.00              97.28                    2.72
## 27   1,732,316           0.00              97.57                    2.43
## 28       1,152           0.00              96.70                    3.30
## 29   9,742,792           2.10              97.63                    2.37
## 30   1,476,427           0.32              97.63                    2.37
## 31   2,693,655           0.58              97.51                    2.49
##    Yield..Mbases. X..PF.Cluster X.....Q30 Mean.Quality X.2 X.3 X.4
## 1           1,363           100     93.82        35.92  NA  NA  NA
## 2           1,957           100     93.85        35.94  NA  NA  NA
## 3           1,353           100     93.95        35.96  NA  NA  NA
## 4             996           100     94.12        35.98  NA  NA  NA
## 5           1,376           100     93.94        35.94  NA  NA  NA
## 6           1,383           100     93.75        35.92  NA  NA  NA
## 7           2,080           100     93.61        35.89  NA  NA  NA
## 8           2,507           100     93.83        35.93  NA  NA  NA
## 9             903           100     93.71        35.90  NA  NA  NA
## 10          1,696           100     93.55        35.88  NA  NA  NA
## 11            887           100     93.75        35.91  NA  NA  NA
## 12          2,900           100     93.93        35.94  NA  NA  NA
## 13          1,324           100     93.90        35.95  NA  NA  NA
## 14          2,217           100     93.93        35.96  NA  NA  NA
## 15            544           100     93.58        35.88  NA  NA  NA
## 16          2,033           100     93.62        35.89  NA  NA  NA
## 17          2,386           100     94.09        35.98  NA  NA  NA
## 18          1,588           100     93.81        35.93  NA  NA  NA
## 19            137           100     92.97        35.76  NA  NA  NA
## 20          2,564           100     93.34        35.84  NA  NA  NA
## 21            516           100     93.06        35.78  NA  NA  NA
## 22          1,191           100     93.55        35.88  NA  NA  NA
## 23          3,299           100     93.51        35.87  NA  NA  NA
## 24          1,533           100     93.77        35.92  NA  NA  NA
## 25          2,040           100     92.53        35.68  NA  NA  NA
## 26            928           100     92.94        35.76  NA  NA  NA
## 27            523           100     93.82        35.92  NA  NA  NA
## 28              0           100     92.06        35.56  NA  NA  NA
## 29          2,942           100     93.00        35.78  NA  NA  NA
## 30            446           100     93.25        35.82  NA  NA  NA
## 31            813           100     93.01        35.79  NA  NA  NA

ssHSA <- ssHSA[ssHSA$Sample %in% colnames(x),]

dim(ssHSA)

## [1] 29 31

x1 <- x[,which(colnames(x) %in% ssHSA$Sample)]
dim(x1)

## [1] 310  29

x1 <- x1[which(rowMeans(x1)>10),]
dim(x1)

## [1] 256  29

ssHSA %>%  kbl(caption="Sample sheet")%>% kable_styling("hover",full_width=FALSE) 

Sample sheet

	Lane	Project	Client.ID	Sample	X	Category	Sex.Age.Season	Sex	Age	Season	Location	Agebyyear	Weight	X.1	Number.of.Tumours	Ulcerated	UlcerationDepth	Secondary.infection	Total.tumour.volume	Barcode.sequence	PF.Clusters	X..of.the.lane	X..Perfect.barcode	X..One.mismatch.barcode	Yield..Mbases.	X..PF.Cluster	X…..Q30	Mean.Quality	X.2	X.3	X.4
1	1	default	S1_LauraGrace	6102461	NA	Healthy/Wild	FAP	F	A	Spring	Channel	2.6219178	7.2	NA	NA	NA		NA	NA	AGAGTAGC+AAGGCGTA	4,511,702	1.00	98.29	1.71	1,363	100	93.82	35.92	NA	NA	NA
2	1	default	S1_Turmeric	6102465	NA	Healthy/Wild	FAU	F	A	Summer	Channel	1.9095890	7.8	NA	NA	NA		NA	NA	GAAGGAAG+GACGTCAT	6,479,047	1.00	97.28	2.72	1,957	100	93.85	35.94	NA	NA	NA
3	1	default	S1_Singapore	6102466	NA	Healthy	FAA	F	A	Autumn	Channel	5.0000000	8.6	NA	NA	NA		NA	NA	ACAGCTCA+TACTGCTC	4,481,213	1.00	97.99	2.01	1,353	100	93.95	35.96	NA	NA	NA
4	1	default	S2_Belinda	6102467	NA	Healthy/Wild	FAU	F	A	Summer	Channel	1.8657534	6.6	NA	NA	NA		NA	NA	GCAAGATC+AGTCGAAG	3,296,547	1.00	97.40	2.60	996	100	94.12	35.98	NA	NA	NA
5	1	default	S2_Rosie	6102470	NA	Healthy/Wild	FAU	F	A	Summer	Channel	1.8904110	7.4	NA	NA	NA		NA	NA	CGCATGAT+AAGCCTGA	4,556,781	1.00	95.70	4.30	1,376	100	93.94	35.94	NA	NA	NA
6	1	default	S3_ST62	6102475	NA	Healthy/Wild	FAW	F	A	Winter	Channel	3.2712329	10.2	NA	NA	NA		NA	NA	TGAGCTAG+GAACGGTT	4,579,681	1.00	96.67	3.33	1,383	100	93.75	35.92	NA	NA	NA
7	1	default	S4_Kerchak	6102478	NA	Healthy	MAA	M	A	Autumn	Channel	4.1863014	11.6	NA	NA	NA		NA	NA	CGGCTAAT+CTCGTTCT	6,886,511	1.00	97.80	2.20	2,080	100	93.61	35.89	NA	NA	NA
8	1	default	S4_Duffman	6102479	NA	Healthy	MAA	M	A	Autumn	Channel	3.1123288	12.4	NA	NA	NA		NA	NA	GACGATCT+TCGTGCAT	8,301,336	2.00	97.75	2.25	2,507	100	93.83	35.93	NA	NA	NA
9	1	default	S4_Hogan	6102481	NA	Healthy	MAA	M	A	Autumn	Channel	3.1616438	11.6	NA	NA	NA		NA	NA	TACGCCTT+GCTACTCT	2,990,345	1.00	97.77	2.23	903	100	93.71	35.90	NA	NA	NA
10	1	default	S5_Silky	6102486	NA	Healthy/Wild	MAW	M	A	Winter	Channel	2.2849315	9.5	NA	NA	NA		NA	NA	CTGTTGAC+ACCTCAGT	5,617,260	1.00	97.81	2.19	1,696	100	93.55	35.88	NA	NA	NA
11	1	default	S5_Longley	6102487	NA	Healthy/Wild	MAU	M	A	Summer	Channel	2.9315068	9.7	NA	NA	NA		NA	NA	CGCAATCT+ACAGGCAT	2,937,660	1.00	97.71	2.29	887	100	93.75	35.91	NA	NA	NA
12	1	default	S6_Myanmar	6102490	NA	Healthy	FAP	F	A	Spring	Channel	5.5726027	7.6	NA	NA	NA		NA	NA	GCACAACT+CTGAACGT	9,604,175	2.00	98.14	1.86	2,900	100	93.93	35.94	NA	NA	NA
13	1	default	S6_Pakana	6102491	NA	Healthy	MSP	M	S	Spring	Channel	1.6712329	7.8	NA	NA	NA		NA	NA	GTTACGCA+ATGGCGAT	4,383,521	1.00	98.09	1.91	1,324	100	93.90	35.95	NA	NA	NA
14	1	default	S6_Hobbit	6102495	NA	Healthy	MSP	M	S	Spring	Channel	1.6273973	9	NA	NA	NA		NA	NA	ATATGCGC+ACGATCAG	7,342,414	2.00	97.00	3.00	2,217	100	93.93	35.96	NA	NA	NA
15	1	default	S7_Lieke	6102499	NA	Healthy/Wild	FSU	F	S	Summer	Channel	0.8657534	4.9	NA	NA	NA		NA	NA	TTCCAAGG+CTACAAGG	1,800,468	0.00	97.79	2.21	544	100	93.58	35.88	NA	NA	NA
16	1	default	S7_Smithers	6102501	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	3.1013699	8.9	NA	NA	NA		NA	NA	GCGTCATT+CAGACGTT	6,731,177	1.00	98.04	1.96	2,033	100	93.62	35.89	NA	NA	NA
17	1	default	S8_Vinnie	6102503	NA	Healthy/Wild	MAU	M	A	Summer	Channel	2.9150685	8.8	NA	NA	NA		NA	NA	CGACGTTA+CTCTGGAT	7,901,958	2.00	98.02	1.98	2,386	100	94.09	35.98	NA	NA	NA
18	1	default	S8_Struggles	6102504	NA	Healthy/Wild	MAW	M	A	Winter	Channel	3.2821918	11.1	NA	NA	NA		NA	NA	ATCGATCG+TGGAAGCA	5,257,820	1.00	95.13	4.87	1,588	100	93.81	35.93	NA	NA	NA
20	1	default	S9_Majula	6102509	NA	Healthy	FAA	F	A	Autumn	Channel	3.1150685	6	NA	NA	NA		NA	NA	ACTCTCGA+CTGTACCA	8,490,270	2.00	97.99	2.01	2,564	100	93.34	35.84	NA	NA	NA
21	1	default	S9_Snoopy	6102510	NA	Healthy/Wild	MSW	M	S	Winter	Channel	1.2794521	6.3	NA	NA	NA		NA	NA	AACTGAGC+CAATCAGG	1,709,078	0.00	98.04	1.96	516	100	93.06	35.78	NA	NA	NA
22	1	default	S9_Sarabi	6102513	NA	Healthy/Wild	FSP	F	S	Spring	Channel	1.6328767	5.8	NA	NA	NA		NA	NA	CTTAGGAC+ACTCCTAC	3,945,235	1.00	98.11	1.89	1,191	100	93.55	35.88	NA	NA	NA
23	1	default	S11_Baloo	6102518	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	3.1726027	11.5	NA	NA	NA		NA	NA	CTGATCGT+GCGCATAT	10,923,019	2.00	98.07	1.93	3,299	100	93.51	35.87	NA	NA	NA
24	1	default	S11_DanRhodes	6102523	NA	Healthy/Wild	MSW	M	S	Winter	Channel	1.3917808	6.8	NA	NA	NA		NA	NA	TCGCTGTT+CAAGTCGT	5,076,673	1.00	97.41	2.59	1,533	100	93.77	35.92	NA	NA	NA
25	1	default	S12_Robin	6102525	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	2.1972603	7.7	NA	NA	NA		NA	NA	ATGACGTC+CTTCCTTC	6,756,342	1.00	97.12	2.88	2,040	100	92.53	35.68	NA	NA	NA
26	1	default	S12_Eeyore	6102527	NA	Healthy/Wild	MAW	M	A	Winter	Channel	3.2794521	9	NA	NA	NA		NA	NA	TCAGGCTT+ATCATGCG	3,071,283	1.00	97.28	2.72	928	100	92.94	35.76	NA	NA	NA
27	1	default	S13_Pangea	6102533	NA	Healthy/Wild	MAU	M	A	Summer	Channel	1.8794521	7.7	NA	NA	NA		NA	NA	ACCACGAT+GTCTGCAA	1,732,316	0.00	97.57	2.43	523	100	93.82	35.92	NA	NA	NA
29	2	default	S22_Ginger	6102586	NA	Healthy/Wild	MSW	M	S	Winter	Channel	1.2712329	7.6	NA	NA	NA		NA	NA	ACAGCTCA+TACTGCTC	9,742,792	2.10	97.63	2.37	2,942	100	93.00	35.78	NA	NA	NA
30	2	default	S24_Boy	6102593	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	3.1452055	8.3	NA	NA	NA		NA	NA	CGACGTTA+CTCTGGAT	1,476,427	0.32	97.63	2.37	446	100	93.25	35.82	NA	NA	NA
31	2	default	S24_Shazza	6102598	NA	Healthy	FAW	F	A	Winter	Channel	2.0273973	6.2	NA	NA	NA		NA	NA	GAGCAGTA+TGAGCTGT	2,693,655	0.58	97.51	2.49	813	100	93.01	35.79	NA	NA	NA

col <- as.character(as.numeric(grepl("A",ssHSA$Age)))
col <- gsub("0","orange",col)
col <- gsub("1","brown",col)

mds<-plotMDS(x1,pch=19,cex=3,col=col,main="MDS plot: Comparison of Sub-Adult vs Adult Healthy devils")
mtext("Sub-Adult=Orange, Adult=Brown")

ssHSA$MvF <- factor(ssHSA$Sex)
ssHSA$SubAdvAd <- as.numeric(factor(ssHSA$Age,levels=c("S","A"),ordered = TRUE))
ssHSA$Season <- factor(ssHSA$Season)
str(ssHSA)

## 'data.frame':    29 obs. of  33 variables:
##  $ Lane                   : int  1 1 1 1 1 1 1 1 1 1 ...
##  $ Project                : chr  "default" "default" "default" "default" ...
##  $ Client.ID              : chr  "S1_LauraGrace" "S1_Turmeric" "S1_Singapore" "S2_Belinda" ...
##  $ Sample                 : int  6102461 6102465 6102466 6102467 6102470 6102475 6102478 6102479 6102481 6102486 ...
##  $ X                      : logi  NA NA NA NA NA NA ...
##  $ Category               : chr  "Healthy/Wild" "Healthy/Wild" "Healthy" "Healthy/Wild" ...
##  $ Sex.Age.Season         : chr  "FAP" "FAU" "FAA" "FAU" ...
##  $ Sex                    : chr  "F" "F" "F" "F" ...
##  $ Age                    : chr  "A" "A" "A" "A" ...
##  $ Season                 : Factor w/ 4 levels "Autumn","Spring",..: 2 3 1 3 3 4 1 1 1 4 ...
##  $ Location               : chr  "Channel" "Channel" "Channel" "Channel" ...
##  $ Agebyyear              : num  2.62 1.91 5 1.87 1.89 ...
##  $ Weight                 : chr  "7.2" "7.8" "8.6" "6.6" ...
##  $ X.1                    : logi  NA NA NA NA NA NA ...
##  $ Number.of.Tumours      : int  NA NA NA NA NA NA NA NA NA NA ...
##  $ Ulcerated              : logi  NA NA NA NA NA NA ...
##  $ UlcerationDepth        : chr  "" "" "" "" ...
##  $ Secondary.infection    : logi  NA NA NA NA NA NA ...
##  $ Total.tumour.volume    : num  NA NA NA NA NA NA NA NA NA NA ...
##  $ Barcode.sequence       : chr  "AGAGTAGC+AAGGCGTA" "GAAGGAAG+GACGTCAT" "ACAGCTCA+TACTGCTC" "GCAAGATC+AGTCGAAG" ...
##  $ PF.Clusters            : chr  "4,511,702" "6,479,047" "4,481,213" "3,296,547" ...
##  $ X..of.the.lane         : num  1 1 1 1 1 1 1 2 1 1 ...
##  $ X..Perfect.barcode     : num  98.3 97.3 98 97.4 95.7 ...
##  $ X..One.mismatch.barcode: num  1.71 2.72 2.01 2.6 4.3 3.33 2.2 2.25 2.23 2.19 ...
##  $ Yield..Mbases.         : chr  "1,363" "1,957" "1,353" "996" ...
##  $ X..PF.Cluster          : int  100 100 100 100 100 100 100 100 100 100 ...
##  $ X.....Q30              : num  93.8 93.8 94 94.1 93.9 ...
##  $ Mean.Quality           : num  35.9 35.9 36 36 35.9 ...
##  $ X.2                    : logi  NA NA NA NA NA NA ...
##  $ X.3                    : logi  NA NA NA NA NA NA ...
##  $ X.4                    : logi  NA NA NA NA NA NA ...
##  $ MvF                    : Factor w/ 2 levels "F","M": 1 1 1 1 1 1 2 2 2 2 ...
##  $ SubAdvAd               : num  2 2 2 2 2 2 2 2 2 2 ...

dds <- DESeqDataSetFromMatrix(countData = x1 , colData = ssHSA, design = ~ Season + MvF + SubAdvAd)

## converting counts to integer mode

##   the design formula contains one or more numeric variables with integer values,
##   specifying a model with increasing fold change for higher values.
##   did you mean for this to be a factor? if so, first convert
##   this variable to a factor using the factor() function

res <- DESeq(dds)

## estimating size factors

## estimating dispersions

## gene-wise dispersion estimates

## mean-dispersion relationship

## final dispersion estimates

## fitting model and testing

## -- replacing outliers and refitting for 10 genes
## -- DESeq argument 'minReplicatesForReplace' = 7 
## -- original counts are preserved in counts(dds)

## estimating dispersions

## fitting model and testing

z<- results(res)
vsd <- vst(dds, blind=FALSE,nsub=nrow(x1)-50)
zz<-cbind(as.data.frame(z),assay(vsd))
dge<-as.data.frame(zz[order(zz$pvalue),])
dgeHSA <- dge
sig <- subset(dge,padj<0.05)
sig1_up <- rownames(subset(sig,log2FoldChange>0))
sig1_dn <- rownames(subset(sig,log2FoldChange<0))
length(sig1_up)

## [1] 5

maplot(dgeHSA,"Sub-Adult vs Adult Healthy devils")

make_volcano(dgeHSA,"Sub-Adult vs Adult Healthy devils")

col <- as.character(as.numeric(grepl("A",ssHSA$Age)))
col <- gsub("0","orange",col)
col <- gsub("1","brown",col)
mtext("Sub-Adult=Orange, Adult=Brown")

sig[1:100,1:6] %>%
  kbl(caption="Comparison of Sub-Adult vs Adult Healthy devils") %>%
  kable_paper("hover", full_width = F)

Comparison of Sub-Adult vs Adult Healthy devils

	baseMean	log2FoldChange	lfcSE	stat	pvalue	padj
CD19	2631.3365	-1.6606475	0.3167409	-5.242920	0.0000002	0.0000319
FCMR	30389.8997	-1.9153371	0.3730840	-5.133796	0.0000003	0.0000319
MS4A1	6756.4847	-1.5648906	0.3200891	-4.888922	0.0000010	0.0000760
IgM	57530.3797	-1.4691006	0.3060463	-4.800256	0.0000016	0.0000891
CXCR5	1116.1701	-1.3356203	0.2825809	-4.726506	0.0000023	0.0001028
CD40	2105.5735	-1.2661072	0.2865620	-4.418266	0.0000099	0.0003731
CD22	8615.2980	-1.2218493	0.2825749	-4.323983	0.0000153	0.0004925
TLR9	642.7367	-1.0914975	0.2626616	-4.155527	0.0000325	0.0009128
IL16	10766.3462	-0.5052720	0.1256632	-4.020845	0.0000580	0.0014497
PDCD1	998.1827	-1.3735597	0.3646301	-3.766995	0.0001652	0.0037176
Saha-UM	151.8341	1.7735634	0.5327878	3.328836	0.0008721	0.0178383
TLR4	1310.8396	1.1932591	0.3682601	3.240262	0.0011942	0.0206801
CCLD13	277.1329	1.7843477	0.5507065	3.240107	0.0011949	0.0206801
IL1B	5176.4086	0.9592865	0.3225449	2.974117	0.0029383	0.0457211
IL3Ra	4282.2086	0.8604834	0.2904244	2.962848	0.0030481	0.0457211
NA	NA	NA	NA	NA	NA	NA
NA.1	NA	NA	NA	NA	NA	NA
NA.2	NA	NA	NA	NA	NA	NA
NA.3	NA	NA	NA	NA	NA	NA
NA.4	NA	NA	NA	NA	NA	NA
NA.5	NA	NA	NA	NA	NA	NA
NA.6	NA	NA	NA	NA	NA	NA
NA.7	NA	NA	NA	NA	NA	NA
NA.8	NA	NA	NA	NA	NA	NA
NA.9	NA	NA	NA	NA	NA	NA
NA.10	NA	NA	NA	NA	NA	NA
NA.11	NA	NA	NA	NA	NA	NA
NA.12	NA	NA	NA	NA	NA	NA
NA.13	NA	NA	NA	NA	NA	NA
NA.14	NA	NA	NA	NA	NA	NA
NA.15	NA	NA	NA	NA	NA	NA
NA.16	NA	NA	NA	NA	NA	NA
NA.17	NA	NA	NA	NA	NA	NA
NA.18	NA	NA	NA	NA	NA	NA
NA.19	NA	NA	NA	NA	NA	NA
NA.20	NA	NA	NA	NA	NA	NA
NA.21	NA	NA	NA	NA	NA	NA
NA.22	NA	NA	NA	NA	NA	NA
NA.23	NA	NA	NA	NA	NA	NA
NA.24	NA	NA	NA	NA	NA	NA
NA.25	NA	NA	NA	NA	NA	NA
NA.26	NA	NA	NA	NA	NA	NA
NA.27	NA	NA	NA	NA	NA	NA
NA.28	NA	NA	NA	NA	NA	NA
NA.29	NA	NA	NA	NA	NA	NA
NA.30	NA	NA	NA	NA	NA	NA
NA.31	NA	NA	NA	NA	NA	NA
NA.32	NA	NA	NA	NA	NA	NA
NA.33	NA	NA	NA	NA	NA	NA
NA.34	NA	NA	NA	NA	NA	NA
NA.35	NA	NA	NA	NA	NA	NA
NA.36	NA	NA	NA	NA	NA	NA
NA.37	NA	NA	NA	NA	NA	NA
NA.38	NA	NA	NA	NA	NA	NA
NA.39	NA	NA	NA	NA	NA	NA
NA.40	NA	NA	NA	NA	NA	NA
NA.41	NA	NA	NA	NA	NA	NA
NA.42	NA	NA	NA	NA	NA	NA
NA.43	NA	NA	NA	NA	NA	NA
NA.44	NA	NA	NA	NA	NA	NA
NA.45	NA	NA	NA	NA	NA	NA
NA.46	NA	NA	NA	NA	NA	NA
NA.47	NA	NA	NA	NA	NA	NA
NA.48	NA	NA	NA	NA	NA	NA
NA.49	NA	NA	NA	NA	NA	NA
NA.50	NA	NA	NA	NA	NA	NA
NA.51	NA	NA	NA	NA	NA	NA
NA.52	NA	NA	NA	NA	NA	NA
NA.53	NA	NA	NA	NA	NA	NA
NA.54	NA	NA	NA	NA	NA	NA
NA.55	NA	NA	NA	NA	NA	NA
NA.56	NA	NA	NA	NA	NA	NA
NA.57	NA	NA	NA	NA	NA	NA
NA.58	NA	NA	NA	NA	NA	NA
NA.59	NA	NA	NA	NA	NA	NA
NA.60	NA	NA	NA	NA	NA	NA
NA.61	NA	NA	NA	NA	NA	NA
NA.62	NA	NA	NA	NA	NA	NA
NA.63	NA	NA	NA	NA	NA	NA
NA.64	NA	NA	NA	NA	NA	NA
NA.65	NA	NA	NA	NA	NA	NA
NA.66	NA	NA	NA	NA	NA	NA
NA.67	NA	NA	NA	NA	NA	NA
NA.68	NA	NA	NA	NA	NA	NA
NA.69	NA	NA	NA	NA	NA	NA
NA.70	NA	NA	NA	NA	NA	NA
NA.71	NA	NA	NA	NA	NA	NA
NA.72	NA	NA	NA	NA	NA	NA
NA.73	NA	NA	NA	NA	NA	NA
NA.74	NA	NA	NA	NA	NA	NA
NA.75	NA	NA	NA	NA	NA	NA
NA.76	NA	NA	NA	NA	NA	NA
NA.77	NA	NA	NA	NA	NA	NA
NA.78	NA	NA	NA	NA	NA	NA
NA.79	NA	NA	NA	NA	NA	NA
NA.80	NA	NA	NA	NA	NA	NA
NA.81	NA	NA	NA	NA	NA	NA
NA.82	NA	NA	NA	NA	NA	NA
NA.83	NA	NA	NA	NA	NA	NA
NA.84	NA	NA	NA	NA	NA	NA

write.table(dge,file="dgeHSA.tsv",sep="\t")

mx <- sig[,7:ncol(sig)]
mx <- head(mx,30)
colfunc <- colorRampPalette(c("blue", "white", "red"))
heatmap.2(as.matrix(mx),trace="none",scale="row",
          col=colfunc(25),ColSideColors=col,mar=c(5,12))

ssDFT <- filter(ss, Category %in% c('DFT1', 'DFT2'))

ssDFTAg <- ssDFT[which(grepl("S",ssDFT$Age) + grepl("A",ssDFT$Age)>0),]
ssDFTAg

##    Lane Project              Client.ID  Sample  X Category Sex.Age.Season Sex
## 1     1 default            S1_Sarafina 6102463 NA     DFT1            FAP   F
## 2     1 default                S1_Lucy 6102464 NA     DFT1            FAU   F
## 3     1 default                S2_Hope 6102471 NA     DFT1            FAU   F
## 4     1 default               S2_Sally 6102472 NA     DFT1            FAU   F
## 5     1 default                 S3_Apu 6102473 NA     DFT1            MAA   M
## 6     1 default              S3_Simone 6102474 NA     DFT1            FAW   F
## 7     1 default              S4_Wiggum 6102480 NA     DFT1            MAA   M
## 8     1 default                S4_Pooh 6102482 NA     DFT1            MAA   M
## 9     1 default            S4_Milhouse 6102483 NA     DFT1            MAA   M
## 10    1 default               S5_Allen 6102488 NA     DFT1            MAU   M
## 11    1 default                S5_Cyan 6102489 NA     DFT1            MAW   M
## 12    1 default              S6_Shelob 6102492 NA     DFT2            FAP   F
## 13    1 default           S6_Stuttgart 6102493 NA     DFT1            MSP   M
## 14    1 default              S7_Violet 6102498 NA     DFT2            FSU   F
## 15    1 default              S7_Willie 6102500 NA     DFT2            MAA   M
## 16    1 default              S8_Orange 6102502 NA     DFT1            FAU   F
## 17    1 default                S8_ST67 6102505 NA     DFT2            MAW   M
## 18    1 default              S8_Yellow 6102507 NA     DFT2            MAU   M
## 19    1 default                S9_Jane 6102508 NA     DFT1            FAA   F
## 20    1 default        S9_Clytemnestra 6102511 NA     DFT1            FSP   F
## 21    1 default                S9_Zeus 6102512 NA     DFT2            MSW   M
## 22    1 default           S10_Laurasia 6102516 NA     DFT2            MSP   M
## 23    1 default          S10_Enchilada 6102517 NA     DFT2            MAP   M
## 24    1 default S11_Murdoch/Washington 6102519 NA     DFT2            MAA   M
## 25    1 default                S12_Abe 6102524 NA     DFT1            MAA   M
## 26    1 default              S12_Lilac 6102526 NA     DFT1            MAW   M
## 27    1 default              S13_Green 6102531 NA     DFT1            MAA   M
## 28    1 default               S13_Bump 6102532 NA     DFT1            MAU   M
## 29    1 default             S14_Purple 6102536 NA     DFT1            FAW   F
##    Age Season Location Agebyyear Weight X.1 Number.of.Tumours Ulcerated
## 1    A Spring  Channel 2.6191781    6.4  NA                NA        NA
## 2    A Summer  Channel 2.9178082    7.1  NA                 3      TRUE
## 3    A Summer  Channel 2.9315068    6.4  NA                 4      TRUE
## 4    A Summer  Channel 2.9945205      7  NA                 1      TRUE
## 5    A Autumn  Channel 3.1150685    9.5  NA                 1     FALSE
## 6    A Winter  Channel 3.2739726   10.2  NA                 1      TRUE
## 7    A Autumn  Channel 3.1178082    8.3  NA                 2     FALSE
## 8    A Autumn  Channel 4.1863014   11.8  NA                 1     FALSE
## 9    A Autumn  Channel 4.1013699     11  NA                 1      TRUE
## 10   A Summer  Channel 1.9095890    6.7  NA                 1      TRUE
## 11   A Winter  Channel 3.3945205    9.1  NA                 5      TRUE
## 12   A Spring  Channel 4.6219178    7.2  NA                NA        NA
## 13   S Spring  Channel 0.5780822   12.7  NA                 1     FALSE
## 14   S Summer  Channel 0.8739726    4.6  NA                 1      TRUE
## 15   A Autumn  Channel 2.1150685    8.2  NA                 2      TRUE
## 16   A Summer  Channel 1.9150685    7.1  NA                 7      TRUE
## 17   A Winter  Channel 4.2794521   10.2  NA                 1      TRUE
## 18   A Summer  Channel 1.8657534   10.5  NA                 2      TRUE
## 19   A Autumn  Channel 3.1972603    6.1  NA                NA        NA
## 20   S Spring  Channel 1.6602740      6  NA                NA        NA
## 21   S Winter  Channel 1.3479452    7.3  NA                NA        NA
## 22   S Spring  Channel 1.6383562      8  NA                NA        NA
## 23   A Spring  Channel 3.6438356    9.3  NA                14     FALSE
## 24   A Autumn  Channel 1.4438356    8.9  NA                NA        NA
## 25   A Autumn  Channel 3.1150685    7.9  NA                 1      TRUE
## 26   A Winter  Channel 3.4109589    9.6  NA                 5      TRUE
## 27   A Autumn  Channel 3.1534247    7.5  NA                 2      TRUE
## 28   A Summer  Channel 3.9068493    9.3  NA                NA        NA
## 29   A Winter  Channel 2.3945205    6.8  NA                 1      TRUE
##    UlcerationDepth Secondary.infection Total.tumour.volume  Barcode.sequence
## 1                                   NA           95140.790 ATCCAGAG+TAACGTCG
## 2             deep                TRUE          700825.170 AAGCACTG+AGGTCAAC
## 3             deep                TRUE          103815.530 AGCACTTC+ACGTCCAA
## 4             deep               FALSE            1588.272 CGTTGCAA+ATAGAGCG
## 5      superficial                TRUE             200.000 CCTTGATC+GATGGAGT
## 6      superficial                TRUE            2670.678 CGGTTGTT+GTGGTATG
## 7                                FALSE            7272.309 GTGAAGTG+GAGCAATC
## 8                                FALSE            7577.856 CTTCGACT+GATCTTGC
## 9             deep                TRUE             480.000 AACCGTTC+CTAGCTCA
## 10            deep                TRUE           55959.750 TTGCAGAC+ATCGTGGT
## 11            deep                TRUE          206006.422 GTAGGAGT+GTCCTAAG
## 12                                  NA            7455.435 GTCATCGA+GATCCACT
## 13                                TRUE            3809.806 AGCGTGTT+CTTCGCAA
## 14     superficial                TRUE            5517.330 CTTGTCGA+CGATGTTC
## 15            deep                TRUE            1527.480 TGGACTCT+TCTTACGG
## 16     superficial                TRUE           43690.461 GTTGACCT+CAGTGCTT
## 17            deep                TRUE          102246.432 TTGGACGT+GAAGTGCT
## 18            deep                TRUE          139389.671 GAAGTTGG+CGAACAAC
## 19                                  NA           64646.036 GTTCTCGT+TGTTCCGT
## 20                                  NA           34738.080 CATGGCTA+CACACATC
## 21                                  NA             100.000 CGTGTGTA+CCAACGAA
## 22                                  NA            2697.152 TACGCTAC+CGTGTGAT
## 23                               FALSE          301301.905 GATTGCTC+CACTTCAC
## 24                                  NA          127176.305 ACTGAGGT+GTCAACAG
## 25            deep                TRUE           13125.000 GTTGTTCG+CCAACTTC
## 26     superficial                TRUE           37293.844 TATCGGTC+TACCGGAT
## 27     superficial                TRUE            7407.222 GTCGAAGA+CCACATTG
## 28                                  NA          629522.362 AGTTCGTC+ACTCGATC
## 29            deep                TRUE           24938.123 CGCTCTAT+TTGCAACG
##    PF.Clusters X..of.the.lane X..Perfect.barcode X..One.mismatch.barcode
## 1    6,491,488              1              74.07                   25.93
## 2    6,121,785              1              98.00                    2.00
## 3    1,971,776              0              97.96                    2.04
## 4    2,429,449              1              97.42                    2.58
## 5    4,939,580              1              96.94                    3.06
## 6    4,015,027              1              95.91                    4.09
## 7    4,201,355              1              97.17                    2.83
## 8    3,821,151              1              97.31                    2.69
## 9   10,492,900              2              98.13                    1.87
## 10   5,288,435              1              96.85                    3.15
## 11   4,138,600              1              98.03                    1.97
## 12   3,184,798              1              98.21                    1.79
## 13   3,056,475              1              97.50                    2.50
## 14   2,768,682              1              98.19                    1.81
## 15   4,539,036              1              98.30                    1.70
## 16   5,072,309              1              97.92                    2.08
## 17   5,476,735              1              95.44                    4.56
## 18   6,469,324              1              97.99                    2.01
## 19  12,489,105              3              97.92                    2.08
## 20   6,141,790              1              98.00                    2.00
## 21     699,953              0              97.64                    2.36
## 22   9,716,148              2              97.81                    2.19
## 23  11,967,844              2              97.82                    2.18
## 24   2,003,509              0              97.82                    2.18
## 25   6,446,467              1              97.60                    2.40
## 26   4,490,241              1              98.15                    1.85
## 27   2,623,695              1              97.38                    2.62
## 28   9,435,053              2              98.16                    1.84
## 29   5,432,182              1              97.81                    2.19
##    Yield..Mbases. X..PF.Cluster X.....Q30 Mean.Quality X.2 X.3 X.4
## 1           1,960           100     93.74        35.92  NA  NA  NA
## 2           1,849           100     93.84        35.93  NA  NA  NA
## 3             595           100     93.62        35.88  NA  NA  NA
## 4             734           100     93.78        35.93  NA  NA  NA
## 5           1,492           100     93.71        35.91  NA  NA  NA
## 6           1,213           100     93.42        35.86  NA  NA  NA
## 7           1,269           100     93.64        35.90  NA  NA  NA
## 8           1,154           100     93.60        35.89  NA  NA  NA
## 9           3,169           100     93.80        35.92  NA  NA  NA
## 10          1,597           100     93.42        35.86  NA  NA  NA
## 11          1,250           100     93.67        35.90  NA  NA  NA
## 12            962           100     93.90        35.94  NA  NA  NA
## 13            923           100     93.95        35.95  NA  NA  NA
## 14            836           100     93.85        35.93  NA  NA  NA
## 15          1,371           100     93.84        35.93  NA  NA  NA
## 16          1,532           100     93.36        35.84  NA  NA  NA
## 17          1,654           100     93.21        35.82  NA  NA  NA
## 18          1,954           100     94.00        35.97  NA  NA  NA
## 19          3,772           100     93.54        35.88  NA  NA  NA
## 20          1,855           100     93.42        35.85  NA  NA  NA
## 21            211           100     93.33        35.84  NA  NA  NA
## 22          2,934           100     93.97        35.95  NA  NA  NA
## 23          3,614           100     93.88        35.94  NA  NA  NA
## 24            605           100     93.39        35.85  NA  NA  NA
## 25          1,947           100     93.65        35.90  NA  NA  NA
## 26          1,356           100     93.42        35.85  NA  NA  NA
## 27            792           100     92.94        35.76  NA  NA  NA
## 28          2,849           100     93.64        35.89  NA  NA  NA
## 29          1,641           100     93.93        35.95  NA  NA  NA

ssDFTAg <- ssDFTAg[ssDFTAg$Sample %in% colnames(x),]

dim(ssDFTAg)

## [1] 28 31

x1 <- x[,which(colnames(x) %in% ssDFTAg$Sample)]
dim(x1)

## [1] 310  28

x1 <- x1[which(rowMeans(x1)>10),]
dim(x1)

## [1] 258  28

ssDFTAg %>%  kbl(caption="Sample sheet")%>% kable_styling("hover",full_width=FALSE) 

Sample sheet

	Lane	Project	Client.ID	Sample	X	Category	Sex.Age.Season	Sex	Age	Season	Location	Agebyyear	Weight	X.1	Number.of.Tumours	Ulcerated	UlcerationDepth	Secondary.infection	Total.tumour.volume	Barcode.sequence	PF.Clusters	X..of.the.lane	X..Perfect.barcode	X..One.mismatch.barcode	Yield..Mbases.	X..PF.Cluster	X…..Q30	Mean.Quality	X.2	X.3	X.4
1	1	default	S1_Sarafina	6102463	NA	DFT1	FAP	F	A	Spring	Channel	2.6191781	6.4	NA	NA	NA		NA	95140.790	ATCCAGAG+TAACGTCG	6,491,488	1	74.07	25.93	1,960	100	93.74	35.92	NA	NA	NA
2	1	default	S1_Lucy	6102464	NA	DFT1	FAU	F	A	Summer	Channel	2.9178082	7.1	NA	3	TRUE	deep	TRUE	700825.170	AAGCACTG+AGGTCAAC	6,121,785	1	98.00	2.00	1,849	100	93.84	35.93	NA	NA	NA
3	1	default	S2_Hope	6102471	NA	DFT1	FAU	F	A	Summer	Channel	2.9315068	6.4	NA	4	TRUE	deep	TRUE	103815.530	AGCACTTC+ACGTCCAA	1,971,776	0	97.96	2.04	595	100	93.62	35.88	NA	NA	NA
4	1	default	S2_Sally	6102472	NA	DFT1	FAU	F	A	Summer	Channel	2.9945205	7	NA	1	TRUE	deep	FALSE	1588.272	CGTTGCAA+ATAGAGCG	2,429,449	1	97.42	2.58	734	100	93.78	35.93	NA	NA	NA
5	1	default	S3_Apu	6102473	NA	DFT1	MAA	M	A	Autumn	Channel	3.1150685	9.5	NA	1	FALSE	superficial	TRUE	200.000	CCTTGATC+GATGGAGT	4,939,580	1	96.94	3.06	1,492	100	93.71	35.91	NA	NA	NA
6	1	default	S3_Simone	6102474	NA	DFT1	FAW	F	A	Winter	Channel	3.2739726	10.2	NA	1	TRUE	superficial	TRUE	2670.678	CGGTTGTT+GTGGTATG	4,015,027	1	95.91	4.09	1,213	100	93.42	35.86	NA	NA	NA
7	1	default	S4_Wiggum	6102480	NA	DFT1	MAA	M	A	Autumn	Channel	3.1178082	8.3	NA	2	FALSE		FALSE	7272.309	GTGAAGTG+GAGCAATC	4,201,355	1	97.17	2.83	1,269	100	93.64	35.90	NA	NA	NA
8	1	default	S4_Pooh	6102482	NA	DFT1	MAA	M	A	Autumn	Channel	4.1863014	11.8	NA	1	FALSE		FALSE	7577.856	CTTCGACT+GATCTTGC	3,821,151	1	97.31	2.69	1,154	100	93.60	35.89	NA	NA	NA
9	1	default	S4_Milhouse	6102483	NA	DFT1	MAA	M	A	Autumn	Channel	4.1013699	11	NA	1	TRUE	deep	TRUE	480.000	AACCGTTC+CTAGCTCA	10,492,900	2	98.13	1.87	3,169	100	93.80	35.92	NA	NA	NA
10	1	default	S5_Allen	6102488	NA	DFT1	MAU	M	A	Summer	Channel	1.9095890	6.7	NA	1	TRUE	deep	TRUE	55959.750	TTGCAGAC+ATCGTGGT	5,288,435	1	96.85	3.15	1,597	100	93.42	35.86	NA	NA	NA
11	1	default	S5_Cyan	6102489	NA	DFT1	MAW	M	A	Winter	Channel	3.3945205	9.1	NA	5	TRUE	deep	TRUE	206006.422	GTAGGAGT+GTCCTAAG	4,138,600	1	98.03	1.97	1,250	100	93.67	35.90	NA	NA	NA
12	1	default	S6_Shelob	6102492	NA	DFT2	FAP	F	A	Spring	Channel	4.6219178	7.2	NA	NA	NA		NA	7455.435	GTCATCGA+GATCCACT	3,184,798	1	98.21	1.79	962	100	93.90	35.94	NA	NA	NA
13	1	default	S6_Stuttgart	6102493	NA	DFT1	MSP	M	S	Spring	Channel	0.5780822	12.7	NA	1	FALSE		TRUE	3809.806	AGCGTGTT+CTTCGCAA	3,056,475	1	97.50	2.50	923	100	93.95	35.95	NA	NA	NA
14	1	default	S7_Violet	6102498	NA	DFT2	FSU	F	S	Summer	Channel	0.8739726	4.6	NA	1	TRUE	superficial	TRUE	5517.330	CTTGTCGA+CGATGTTC	2,768,682	1	98.19	1.81	836	100	93.85	35.93	NA	NA	NA
15	1	default	S7_Willie	6102500	NA	DFT2	MAA	M	A	Autumn	Channel	2.1150685	8.2	NA	2	TRUE	deep	TRUE	1527.480	TGGACTCT+TCTTACGG	4,539,036	1	98.30	1.70	1,371	100	93.84	35.93	NA	NA	NA
16	1	default	S8_Orange	6102502	NA	DFT1	FAU	F	A	Summer	Channel	1.9150685	7.1	NA	7	TRUE	superficial	TRUE	43690.461	GTTGACCT+CAGTGCTT	5,072,309	1	97.92	2.08	1,532	100	93.36	35.84	NA	NA	NA
17	1	default	S8_ST67	6102505	NA	DFT2	MAW	M	A	Winter	Channel	4.2794521	10.2	NA	1	TRUE	deep	TRUE	102246.432	TTGGACGT+GAAGTGCT	5,476,735	1	95.44	4.56	1,654	100	93.21	35.82	NA	NA	NA
18	1	default	S8_Yellow	6102507	NA	DFT2	MAU	M	A	Summer	Channel	1.8657534	10.5	NA	2	TRUE	deep	TRUE	139389.671	GAAGTTGG+CGAACAAC	6,469,324	1	97.99	2.01	1,954	100	94.00	35.97	NA	NA	NA
19	1	default	S9_Jane	6102508	NA	DFT1	FAA	F	A	Autumn	Channel	3.1972603	6.1	NA	NA	NA		NA	64646.036	GTTCTCGT+TGTTCCGT	12,489,105	3	97.92	2.08	3,772	100	93.54	35.88	NA	NA	NA
20	1	default	S9_Clytemnestra	6102511	NA	DFT1	FSP	F	S	Spring	Channel	1.6602740	6	NA	NA	NA		NA	34738.080	CATGGCTA+CACACATC	6,141,790	1	98.00	2.00	1,855	100	93.42	35.85	NA	NA	NA
22	1	default	S10_Laurasia	6102516	NA	DFT2	MSP	M	S	Spring	Channel	1.6383562	8	NA	NA	NA		NA	2697.152	TACGCTAC+CGTGTGAT	9,716,148	2	97.81	2.19	2,934	100	93.97	35.95	NA	NA	NA
23	1	default	S10_Enchilada	6102517	NA	DFT2	MAP	M	A	Spring	Channel	3.6438356	9.3	NA	14	FALSE		FALSE	301301.905	GATTGCTC+CACTTCAC	11,967,844	2	97.82	2.18	3,614	100	93.88	35.94	NA	NA	NA
24	1	default	S11_Murdoch/Washington	6102519	NA	DFT2	MAA	M	A	Autumn	Channel	1.4438356	8.9	NA	NA	NA		NA	127176.305	ACTGAGGT+GTCAACAG	2,003,509	0	97.82	2.18	605	100	93.39	35.85	NA	NA	NA
25	1	default	S12_Abe	6102524	NA	DFT1	MAA	M	A	Autumn	Channel	3.1150685	7.9	NA	1	TRUE	deep	TRUE	13125.000	GTTGTTCG+CCAACTTC	6,446,467	1	97.60	2.40	1,947	100	93.65	35.90	NA	NA	NA
26	1	default	S12_Lilac	6102526	NA	DFT1	MAW	M	A	Winter	Channel	3.4109589	9.6	NA	5	TRUE	superficial	TRUE	37293.844	TATCGGTC+TACCGGAT	4,490,241	1	98.15	1.85	1,356	100	93.42	35.85	NA	NA	NA
27	1	default	S13_Green	6102531	NA	DFT1	MAA	M	A	Autumn	Channel	3.1534247	7.5	NA	2	TRUE	superficial	TRUE	7407.222	GTCGAAGA+CCACATTG	2,623,695	1	97.38	2.62	792	100	92.94	35.76	NA	NA	NA
28	1	default	S13_Bump	6102532	NA	DFT1	MAU	M	A	Summer	Channel	3.9068493	9.3	NA	NA	NA		NA	629522.362	AGTTCGTC+ACTCGATC	9,435,053	2	98.16	1.84	2,849	100	93.64	35.89	NA	NA	NA
29	1	default	S14_Purple	6102536	NA	DFT1	FAW	F	A	Winter	Channel	2.3945205	6.8	NA	1	TRUE	deep	TRUE	24938.123	CGCTCTAT+TTGCAACG	5,432,182	1	97.81	2.19	1,641	100	93.93	35.95	NA	NA	NA

col <- as.character(as.numeric(grepl("A",ssDFTAg$Age)))
col <- gsub("0","orange",col)
col <- gsub("1","brown",col)

mds<-plotMDS(x1,pch=19,cex=3,col=col,main="MDS plot: Comparison of Sub-Adult vs Adult DFTD infected devils")
mtext("Sub-Adult=Orange, Adult=Brown")

ssDFTAg$DFTDStat <- grepl("DFT2",ssDFTAg$Category)
ssDFTAg$MvF <- factor(ssDFTAg$Sex)
ssDFTAg$SubAdvAd <- as.numeric(factor(ssDFTAg$Age,levels=c("S","A"),ordered = TRUE))
ssDFTAg$Season <- factor(ssDFTAg$Season)
str(ssDFTAg)

## 'data.frame':    28 obs. of  34 variables:
##  $ Lane                   : int  1 1 1 1 1 1 1 1 1 1 ...
##  $ Project                : chr  "default" "default" "default" "default" ...
##  $ Client.ID              : chr  "S1_Sarafina" "S1_Lucy" "S2_Hope" "S2_Sally" ...
##  $ Sample                 : int  6102463 6102464 6102471 6102472 6102473 6102474 6102480 6102482 6102483 6102488 ...
##  $ X                      : logi  NA NA NA NA NA NA ...
##  $ Category               : chr  "DFT1" "DFT1" "DFT1" "DFT1" ...
##  $ Sex.Age.Season         : chr  "FAP" "FAU" "FAU" "FAU" ...
##  $ Sex                    : chr  "F" "F" "F" "F" ...
##  $ Age                    : chr  "A" "A" "A" "A" ...
##  $ Season                 : Factor w/ 4 levels "Autumn","Spring",..: 2 3 3 3 1 4 1 1 1 3 ...
##  $ Location               : chr  "Channel" "Channel" "Channel" "Channel" ...
##  $ Agebyyear              : num  2.62 2.92 2.93 2.99 3.12 ...
##  $ Weight                 : chr  "6.4" "7.1" "6.4" "7" ...
##  $ X.1                    : logi  NA NA NA NA NA NA ...
##  $ Number.of.Tumours      : int  NA 3 4 1 1 1 2 1 1 1 ...
##  $ Ulcerated              : logi  NA TRUE TRUE TRUE FALSE TRUE ...
##  $ UlcerationDepth        : chr  "" "deep" "deep" "deep" ...
##  $ Secondary.infection    : logi  NA TRUE TRUE FALSE TRUE TRUE ...
##  $ Total.tumour.volume    : num  95141 700825 103816 1588 200 ...
##  $ Barcode.sequence       : chr  "ATCCAGAG+TAACGTCG" "AAGCACTG+AGGTCAAC" "AGCACTTC+ACGTCCAA" "CGTTGCAA+ATAGAGCG" ...
##  $ PF.Clusters            : chr  "6,491,488" "6,121,785" "1,971,776" "2,429,449" ...
##  $ X..of.the.lane         : num  1 1 0 1 1 1 1 1 2 1 ...
##  $ X..Perfect.barcode     : num  74.1 98 98 97.4 96.9 ...
##  $ X..One.mismatch.barcode: num  25.93 2 2.04 2.58 3.06 ...
##  $ Yield..Mbases.         : chr  "1,960" "1,849" "595" "734" ...
##  $ X..PF.Cluster          : int  100 100 100 100 100 100 100 100 100 100 ...
##  $ X.....Q30              : num  93.7 93.8 93.6 93.8 93.7 ...
##  $ Mean.Quality           : num  35.9 35.9 35.9 35.9 35.9 ...
##  $ X.2                    : logi  NA NA NA NA NA NA ...
##  $ X.3                    : logi  NA NA NA NA NA NA ...
##  $ X.4                    : logi  NA NA NA NA NA NA ...
##  $ DFTDStat               : logi  FALSE FALSE FALSE FALSE FALSE FALSE ...
##  $ MvF                    : Factor w/ 2 levels "F","M": 1 1 1 1 2 1 2 2 2 2 ...
##  $ SubAdvAd               : num  2 2 2 2 2 2 2 2 2 2 ...

dds <- DESeqDataSetFromMatrix(countData = x1 , colData = ssDFTAg, design = ~ Season + DFTDStat + MvF + SubAdvAd )

## converting counts to integer mode

##   the design formula contains one or more numeric variables with integer values,
##   specifying a model with increasing fold change for higher values.
##   did you mean for this to be a factor? if so, first convert
##   this variable to a factor using the factor() function

res <- DESeq(dds)

## estimating size factors

## estimating dispersions

## gene-wise dispersion estimates

## mean-dispersion relationship

## final dispersion estimates

## fitting model and testing

z<- results(res)
vsd <- vst(dds, blind=FALSE,nsub=nrow(x1)-50)
zz<-cbind(as.data.frame(z),assay(vsd))
dge<-as.data.frame(zz[order(zz$pvalue),])
dgeDFTAg <- dge
sig <- subset(dge,padj<0.05)
sig1_up <- rownames(subset(sig,log2FoldChange>0))
sig1_dn <- rownames(subset(sig,log2FoldChange<0))
length(sig1_up)

## [1] 5

maplot(dgeDFTAg,"Sub-Adult vs Adult DFTD infected devils")

make_volcano(dgeDFTAg,"Sub-Adult vs Adult DFTD infected devils")

col <- as.character(as.numeric(grepl("A",ssDFTAg$Age)))
col <- gsub("0","brown",col)
col <- gsub("1","orange",col)

sig[1:100,1:6] %>%
  kbl(caption="Comparison of Sub-Adult vs Adult DFTD infected devils") %>%
  kable_paper("hover", full_width = F)

Comparison of Sub-Adult vs Adult DFTD infected devils

	baseMean	log2FoldChange	lfcSE	stat	pvalue	padj
CXCL10	6.852306	14.595049	2.1152809	6.899816	0.0000000	0.0000000
Saha-UM	127.913054	3.351159	0.7142462	4.691882	0.0000027	0.0003492
CD96	100.034988	4.813349	1.0872594	4.427047	0.0000096	0.0008216
XCR1	44.462082	3.136733	0.9280398	3.379955	0.0007250	0.0467047
FCGR2B	980.116968	2.200527	0.6714226	3.277410	0.0010476	0.0467047
IgE	5611.165649	-2.094462	0.6410565	-3.267204	0.0010862	0.0467047
NA	NA	NA	NA	NA	NA	NA
NA.1	NA	NA	NA	NA	NA	NA
NA.2	NA	NA	NA	NA	NA	NA
NA.3	NA	NA	NA	NA	NA	NA
NA.4	NA	NA	NA	NA	NA	NA
NA.5	NA	NA	NA	NA	NA	NA
NA.6	NA	NA	NA	NA	NA	NA
NA.7	NA	NA	NA	NA	NA	NA
NA.8	NA	NA	NA	NA	NA	NA
NA.9	NA	NA	NA	NA	NA	NA
NA.10	NA	NA	NA	NA	NA	NA
NA.11	NA	NA	NA	NA	NA	NA
NA.12	NA	NA	NA	NA	NA	NA
NA.13	NA	NA	NA	NA	NA	NA
NA.14	NA	NA	NA	NA	NA	NA
NA.15	NA	NA	NA	NA	NA	NA
NA.16	NA	NA	NA	NA	NA	NA
NA.17	NA	NA	NA	NA	NA	NA
NA.18	NA	NA	NA	NA	NA	NA
NA.19	NA	NA	NA	NA	NA	NA
NA.20	NA	NA	NA	NA	NA	NA
NA.21	NA	NA	NA	NA	NA	NA
NA.22	NA	NA	NA	NA	NA	NA
NA.23	NA	NA	NA	NA	NA	NA
NA.24	NA	NA	NA	NA	NA	NA
NA.25	NA	NA	NA	NA	NA	NA
NA.26	NA	NA	NA	NA	NA	NA
NA.27	NA	NA	NA	NA	NA	NA
NA.28	NA	NA	NA	NA	NA	NA
NA.29	NA	NA	NA	NA	NA	NA
NA.30	NA	NA	NA	NA	NA	NA
NA.31	NA	NA	NA	NA	NA	NA
NA.32	NA	NA	NA	NA	NA	NA
NA.33	NA	NA	NA	NA	NA	NA
NA.34	NA	NA	NA	NA	NA	NA
NA.35	NA	NA	NA	NA	NA	NA
NA.36	NA	NA	NA	NA	NA	NA
NA.37	NA	NA	NA	NA	NA	NA
NA.38	NA	NA	NA	NA	NA	NA
NA.39	NA	NA	NA	NA	NA	NA
NA.40	NA	NA	NA	NA	NA	NA
NA.41	NA	NA	NA	NA	NA	NA
NA.42	NA	NA	NA	NA	NA	NA
NA.43	NA	NA	NA	NA	NA	NA
NA.44	NA	NA	NA	NA	NA	NA
NA.45	NA	NA	NA	NA	NA	NA
NA.46	NA	NA	NA	NA	NA	NA
NA.47	NA	NA	NA	NA	NA	NA
NA.48	NA	NA	NA	NA	NA	NA
NA.49	NA	NA	NA	NA	NA	NA
NA.50	NA	NA	NA	NA	NA	NA
NA.51	NA	NA	NA	NA	NA	NA
NA.52	NA	NA	NA	NA	NA	NA
NA.53	NA	NA	NA	NA	NA	NA
NA.54	NA	NA	NA	NA	NA	NA
NA.55	NA	NA	NA	NA	NA	NA
NA.56	NA	NA	NA	NA	NA	NA
NA.57	NA	NA	NA	NA	NA	NA
NA.58	NA	NA	NA	NA	NA	NA
NA.59	NA	NA	NA	NA	NA	NA
NA.60	NA	NA	NA	NA	NA	NA
NA.61	NA	NA	NA	NA	NA	NA
NA.62	NA	NA	NA	NA	NA	NA
NA.63	NA	NA	NA	NA	NA	NA
NA.64	NA	NA	NA	NA	NA	NA
NA.65	NA	NA	NA	NA	NA	NA
NA.66	NA	NA	NA	NA	NA	NA
NA.67	NA	NA	NA	NA	NA	NA
NA.68	NA	NA	NA	NA	NA	NA
NA.69	NA	NA	NA	NA	NA	NA
NA.70	NA	NA	NA	NA	NA	NA
NA.71	NA	NA	NA	NA	NA	NA
NA.72	NA	NA	NA	NA	NA	NA
NA.73	NA	NA	NA	NA	NA	NA
NA.74	NA	NA	NA	NA	NA	NA
NA.75	NA	NA	NA	NA	NA	NA
NA.76	NA	NA	NA	NA	NA	NA
NA.77	NA	NA	NA	NA	NA	NA
NA.78	NA	NA	NA	NA	NA	NA
NA.79	NA	NA	NA	NA	NA	NA
NA.80	NA	NA	NA	NA	NA	NA
NA.81	NA	NA	NA	NA	NA	NA
NA.82	NA	NA	NA	NA	NA	NA
NA.83	NA	NA	NA	NA	NA	NA
NA.84	NA	NA	NA	NA	NA	NA
NA.85	NA	NA	NA	NA	NA	NA
NA.86	NA	NA	NA	NA	NA	NA
NA.87	NA	NA	NA	NA	NA	NA
NA.88	NA	NA	NA	NA	NA	NA
NA.89	NA	NA	NA	NA	NA	NA
NA.90	NA	NA	NA	NA	NA	NA
NA.91	NA	NA	NA	NA	NA	NA
NA.92	NA	NA	NA	NA	NA	NA
NA.93	NA	NA	NA	NA	NA	NA

write.table(dge,file="dgeDFTAg.tsv",sep="\t")

ssCap <- filter(ss, Category %in% c('Captive'))

ssCapAg <- ssCap[which(grepl("S",ssCap$Age) + grepl("A",ssCap$Age)>0),]
ssCapAg

##    Lane Project      Client.ID  Sample  X Category Sex.Age.Season Sex Age
## 1     1 default      S2_B99008 6102468 NA  Captive            FAU   F   A
## 2     1 default      S2_Bonney 6102469 NA  Captive            FAU   F   A
## 3     1 default      S3_Jessie 6102476 NA  Captive            FAW   F   A
## 4     1 default    S3_McCauley 6102477 NA  Captive            MAA   M   A
## 5     1 default      S5_Fenton 6102484 NA  Captive            MAU   M   A
## 6     1 default S5_Bryan Adams 6102485 NA  Captive            MAW   M   A
## 7     1 default       S7_Fredo 6102496 NA  Captive            MAA   M   A
## 8     1 default      S7_Piglet 6102497 NA  Captive            FSU   F   S
## 9     1 default    S10_McMahon 6102514 NA  Captive            MSP   M   S
## 10    1 default      S10_Logan 6102515 NA  Captive            MAU   M   A
## 11    1 default       S11_Xuki 6102520 NA  Captive            MAA   M   A
## 12    1 default   S11_Kalamata 6102521 NA  Captive            MSW   M   S
## 13    1 default      S12_Wally 6102528 NA  Captive            MAA   M   A
## 14    1 default        S12_Naz 6102529 NA  Captive            MAW   M   A
## 15    1 default     S13_Humbug 6102530 NA  Captive            MAU   M   A
## 16    1 default     S13_Xerxes 6102535 NA  Captive            MAA   M   A
## 17    1 default     S14_Catani 6102538 NA  Captive            FAU   F   A
## 18    2 default     S15_Amelia 6102541 NA  Captive            FAU   F   A
## 19    2 default     S15_London 6102542 NA  Captive            FAU   F   A
## 20    2 default      S15_Elyse 6102543 NA  Captive            FAP   F   A
## 21    2 default     S16_Marley 6102547 NA  Captive            FAU   F   A
## 22    2 default      S16_Milla 6102548 NA  Captive            FAU   F   A
## 23    2 default S16_Maleficent 6102549 NA  Captive            FAU   F   A
## 24    2 default     S17_Anirak 6102553 NA  Captive            FSW   F   S
## 25    2 default       S17_Poly 6102554 NA  Captive            FSU   F   S
## 26    2 default     S17_Methyl 6102556 NA  Captive            FSU   F   S
## 27    2 default       S18_Zara 6102558 NA  Captive            FSW   F   S
## 28    2 default       S18_Maze 6102559 NA  Captive            FSW   F   S
## 29    2 default    S18_Delores 6102560 NA  Captive            FSW   F   S
## 30    2 default    S19_Morocco 6102564 NA  Captive            MAU   M   A
## 31    2 default   S19_Pecorino 6102565 NA  Captive            FSW   F   S
## 32    2 default      S19_Hooch 6102566 NA  Captive            MSA   M   S
## 33    2 default      S20_Muasy 6102569 NA  Captive            MSU   M   S
## 34    2 default       S20_Fuse 6102570 NA  Captive            MSU   M   S
## 35    2 default     S20_Joseph 6102571 NA  Captive            MSU   M   S
## 36    2 default      S21_Moama 6102576 NA  Captive            MAW   M   A
## 37    2 default     S21_Nauset 6102577 NA  Captive            MSW   M   S
## 38    2 default     S21_Graham 6102578 NA  Captive            MSW   M   S
## 39    2 default      S22_Yanni 6102581 NA  Captive            MSW   M   S
## 40    2 default    S22_Gherkin 6102582 NA  Captive            MSW   M   S
## 41    2 default    S22_Stilton 6102584 NA  Captive            MSW   M   S
## 42    2 default     S23_Osiris 6102588 NA  Captive            MSW   M   S
## 43    2 default     S23_X-boxx 6102589 NA  Captive            FSW   F   S
## 44    2 default     S23_Quince 6102590 NA  Captive            MSW   M   S
## 45    2 default     S24_Malfoy 6102595 NA  Captive            MAW   M   A
## 46    2 default   S24_Licorice 6102596 NA  Captive            FSA   F   S
## 47    2 default       S25_Tofu 6102599 NA  Captive            FAU   F   A
## 48    2 default    S25_Colette 6102600 NA  Captive            FAP   F   A
## 49    2 default S25_CannonBall 6102601 NA  Captive            MSW   M   S
## 50    2 default  S25_Cornelius 6102602 NA  Captive            MAU   M   A
## 51    2 default    S25_Mardook 6102603 NA  Captive            MAW   M   A
## 52    2 default    S25_Negroni 6102604 NA  Captive            FSP   F   S
##    Season    Location Agebyyear Weight X.1 Number.of.Tumours Ulcerated
## 1  Summer     Monarto 3.7369863   8.12  NA                NA        NA
## 2  Summer Healesville 2.5863014    5.4  NA                NA        NA
## 3  Winter     Monarto 2.1945205  12.95  NA                NA        NA
## 4  Autumn Healesville 3.0328767   8.52  NA                NA        NA
## 5  Summer     Monarto 3.0000000    N/A  NA                NA        NA
## 6  Winter Healesville 4.1616438    8.6  NA                NA        NA
## 7  Autumn Healesville 2.9945205    8.8  NA                NA        NA
## 8  Summer Healesville 1.5890411   4.76  NA                NA        NA
## 9  Spring Healesville 1.6493151   8.06  NA                NA        NA
## 10 Summer Healesville 5.6000000   9.02  NA                NA        NA
## 11 Autumn     Monarto 2.0767123   7.95  NA                NA        NA
## 12 Winter Healesville 1.2849315   5.58  NA                NA        NA
## 13 Autumn     Monarto 2.1616438    N/A  NA                NA        NA
## 14 Winter     Monarto 4.4547945    N/A  NA                NA        NA
## 15 Summer     Monarto 2.5397260   7.94  NA                NA        NA
## 16 Autumn     Monarto 2.0767123    N/A  NA                NA        NA
## 17 Summer Healesville 2.8630137   6.46  NA                NA        NA
## 18 Summer Healesville 4.7260274   5.85  NA                NA        NA
## 19 Summer Healesville 3.7863014    4.8  NA                NA        NA
## 20 Spring Healesville 4.5232877   7.88  NA                NA        NA
## 21 Summer     Monarto 4.7726027   7.51  NA                NA        NA
## 22 Summer Healesville 2.7671233   5.82  NA                NA        NA
## 23 Summer     Monarto 4.5589041   8.59  NA                NA        NA
## 24 Winter Healesville 1.3424658   5.42  NA                NA        NA
## 25 Summer     Monarto 1.7917808   6.67  NA                NA        NA
## 26 Summer Healesville 0.7534247   2.72  NA                NA        NA
## 27 Winter     Monarto 1.5150685    4.6  NA                NA        NA
## 28 Winter Healesville 1.4000000   5.82  NA                NA        NA
## 29 Winter Healesville 1.2986301    5.1  NA                NA        NA
## 30 Summer     Monarto 4.5917808   9.82  NA                NA        NA
## 31 Winter Healesville 1.1123288   6.02  NA                NA        NA
## 32 Autumn Healesville 1.0164384    6.1  NA                NA        NA
## 33 Summer Healesville 1.4219178   6.62  NA                NA        NA
## 34 Summer Healesville 1.7315068   6.26  NA                NA        NA
## 35 Summer Healesville 1.7479452   7.42  NA                NA        NA
## 36 Winter     Monarto 1.6986301   8.75  NA                NA        NA
## 37 Winter Healesville 1.3041096  4.036  NA                NA        NA
## 38 Winter Healesville 1.3616438    6.4  NA                NA        NA
## 39 Winter     Monarto 1.5178082   8.15  NA                NA        NA
## 40 Winter Healesville 1.2630137   6.32  NA                NA        NA
## 41 Winter Healesville 1.2986301   5.76  NA                NA        NA
## 42 Winter Healesville 1.1753425   5.34  NA                NA        NA
## 43 Winter     Monarto 1.4657534   4.34  NA                NA        NA
## 44 Winter Healesville 1.3835616    N/A  NA                NA        NA
## 45 Winter Healesville 5.1041096   8.02  NA                NA        NA
## 46 Autumn Healesville 1.8246575   5.94  NA                NA        NA
## 47 Summer Healesville 2.7452055    N/A  NA                NA        NA
## 48 Spring Healesville 6.6273973   7.52  NA                NA        NA
## 49 Winter Healesville 1.1479452   6.54  NA                NA        NA
## 50 Summer Healesville 3.6904110   7.74  NA                NA        NA
## 51 Winter Healesville 4.3013699   9.16  NA                NA        NA
## 52 Spring Healesville 1.6767123   5.86  NA                NA        NA
##    UlcerationDepth Secondary.infection Total.tumour.volume  Barcode.sequence
## 1                                   NA                  NA GCTATCCT+AACAGGTG
## 2                                   NA                  NA ACGTTCAG+AGTTGTGC
## 3                                   NA                  NA CCGTAAGA+AGAGTCCA
## 4                                   NA                  NA CAACACCT+AGCTACCA
## 5                                   NA                  NA GATTACCG+TTGGACTG
## 6                                   NA                  NA CACTAGCT+TCGGATTC
## 7                                   NA                  NA AGTGGATC+TCGATGAC
## 8                                   NA                  NA ATCACACG+GTAGCGTA
## 9                                   NA                  NA AACGTCTG+AATGACGC
## 10                                  NA                  NA GTGCCATA+CTCCTAGT
## 11                                  NA                  NA ATCGCCAT+TGCGTAAC
## 12                                  NA                  NA GAACATCG+TCGACAAG
## 13                                  NA                  NA ACGACTTG+AACAGCGA
## 14                                  NA                  NA CCTTGTAG+CCTTGGAA
## 15                                  NA                  NA GAGCAGTA+TGAGCTGT
## 16                                  NA                  NA GATAGGCT+CAGTCACA
## 17                                  NA                  NA TGAAGACG+ACATGCCA
## 18                                  NA                  NA AACGTCTG+AATGACGC
## 19                                  NA                  NA GTCATCGA+GATCCACT
## 20                                  NA                  NA AGCGTGTT+CTTCGCAA
## 21                                  NA                  NA GCAAGATC+AGTCGAAG
## 22                                  NA                  NA TTGCGAAG+AACACGCT
## 23                                  NA                  NA TGCTTCCA+CGATCGAT
## 24                                  NA                  NA TGAGCTAG+GAACGGTT
## 25                                  NA                  NA CTTAGGAC+ACTCCTAC
## 26                                  NA                  NA TGGTACAG+TCGAGAGT
## 27                                  NA                  NA ACCACGAT+GTCTGCAA
## 28                                  NA                  NA ACTGAGGT+GTCAACAG
## 29                                  NA                  NA ACGACTTG+AACAGCGA
## 30                                  NA                  NA TCAGGCTT+ATCATGCG
## 31                                  NA                  NA ATGCCTGT+AGATTGCG
## 32                                  NA                  NA CCGTAAGA+AGAGTCCA
## 33                                  NA                  NA TTCCAAGG+CTACAAGG
## 34                                  NA                  NA GTAGGAGT+GTCCTAAG
## 35                                  NA                  NA CTTCGACT+GATCTTGC
## 36                                  NA                  NA TTGGACGT+GAAGTGCT
## 37                                  NA                  NA TATCGGTC+TACCGGAT
## 38                                  NA                  NA TACGCCTT+GCTACTCT
## 39                                  NA                  NA CGTGTGTA+CCAACGAA
## 40                                  NA                  NA CCTTGTAG+CCTTGGAA
## 41                                  NA                  NA CGCAATCT+ACAGGCAT
## 42                                  NA                  NA AGTGGATC+TCGATGAC
## 43                                  NA                  NA CTGTTGAC+ACCTCAGT
## 44                                  NA                  NA AACCGTTC+CTAGCTCA
## 45                                  NA                  NA TAACCGGT+ATCGTCTC
## 46                                  NA                  NA TTGCAGAC+ATCGTGGT
## 47                                  NA                  NA GAGACGAT+ACCGGTTA
## 48                                  NA                  NA CTTGTCGA+CGATGTTC
## 49                                  NA                  NA TACGCTAC+CGTGTGAT
## 50                                  NA                  NA GAATCCGA+AGCTAGTG
## 51                                  NA                  NA GCACAACT+CTGAACGT
## 52                                  NA                  NA TGTGACTG+AGCCTATC
##    PF.Clusters X..of.the.lane X..Perfect.barcode X..One.mismatch.barcode
## 1    5,338,890           1.00              98.08                    1.92
## 2   14,968,459           3.00              97.82                    2.18
## 3    4,340,498           1.00              97.95                    2.05
## 4   16,944,152           3.00              97.98                    2.02
## 5    5,125,566           1.00              96.21                    3.79
## 6    5,317,767           1.00              97.33                    2.67
## 7    2,690,632           1.00              97.77                    2.23
## 8   12,134,750           2.00              96.94                    3.06
## 9    3,809,607           1.00              96.73                    3.27
## 10   9,606,720           2.00              97.52                    2.48
## 11   6,018,884           1.00              97.88                    2.12
## 12   6,861,544           1.00              97.76                    2.24
## 13   3,000,501           1.00              97.96                    2.04
## 14   5,927,942           1.00              97.40                    2.60
## 15   1,702,608           0.00              98.07                    1.93
## 16   7,628,257           2.00              98.34                    1.66
## 17  13,381,058           3.00              96.98                    3.02
## 18   8,041,818           1.73              96.32                    3.68
## 19   5,758,111           1.24              96.72                    3.28
## 20   5,774,637           1.24              96.76                    3.24
## 21     348,621           0.08              97.24                    2.76
## 22   6,376,997           1.37              97.60                    2.40
## 23   9,181,668           1.98              98.08                    1.92
## 24   3,740,878           0.81              97.15                    2.85
## 25  11,133,566           2.40              97.85                    2.15
## 26   7,887,375           1.70              97.57                    2.43
## 27   7,308,907           1.58              97.70                    2.30
## 28   2,984,930           0.64              97.59                    2.41
## 29          98           0.00              94.90                    5.10
## 30   4,879,629           1.05              97.38                    2.62
## 31   4,857,395           1.05              97.58                    2.42
## 32   1,074,345           0.23              97.56                    2.44
## 33  10,669,237           2.30              97.64                    2.36
## 34  10,603,179           2.29              97.34                    2.66
## 35   5,642,025           1.22              96.37                    3.63
## 36  10,821,895           2.33              96.02                    3.98
## 37   8,693,920           1.87              98.13                    1.87
## 38   3,422,010           0.74              97.72                    2.28
## 39   4,811,453           1.04              97.67                    2.33
## 40   5,913,586           1.27              97.05                    2.95
## 41   8,306,537           1.79              96.94                    3.06
## 42   3,528,471           0.76              97.24                    2.76
## 43          23           0.00             100.00                     NaN
## 44   7,247,793           1.56              97.63                    2.37
## 45   6,908,416           1.49              97.09                    2.91
## 46   4,475,461           0.96              95.05                    4.95
## 47  12,024,421           2.59              95.37                    4.63
## 48  10,751,797           2.32              97.49                    2.51
## 49  13,015,081           2.81              97.36                    2.64
## 50   3,336,983           0.72              97.10                    2.90
## 51   6,826,313           1.47              97.66                    2.34
## 52   4,566,180           0.98              97.68                    2.32
##    Yield..Mbases. X..PF.Cluster X.....Q30 Mean.Quality X.2 X.3 X.4
## 1           1,612           100     93.69        35.91  NA  NA  NA
## 2           4,520           100     93.68        35.91  NA  NA  NA
## 3           1,311           100     93.54        35.88  NA  NA  NA
## 4           5,117           100     93.05        35.78  NA  NA  NA
## 5           1,548           100     93.89        35.94  NA  NA  NA
## 6           1,606           100     93.24        35.83  NA  NA  NA
## 7             813           100     93.77        35.93  NA  NA  NA
## 8           3,665           100     93.34        35.84  NA  NA  NA
## 9           1,151           100     93.07        35.79  NA  NA  NA
## 10          2,901           100     93.25        35.83  NA  NA  NA
## 11          1,818           100     93.58        35.89  NA  NA  NA
## 12          2,072           100     93.49        35.87  NA  NA  NA
## 13            906           100     93.70        35.91  NA  NA  NA
## 14          1,790           100     93.86        35.94  NA  NA  NA
## 15            514           100     93.58        35.89  NA  NA  NA
## 16          2,304           100     93.62        35.89  NA  NA  NA
## 17          4,041           100     93.36        35.84  NA  NA  NA
## 18          2,429           100     92.18        35.63  NA  NA  NA
## 19          1,739           100     91.87        35.55  NA  NA  NA
## 20          1,744           100     91.56        35.50  NA  NA  NA
## 21            105           100     92.77        35.73  NA  NA  NA
## 22          1,926           100     92.44        35.67  NA  NA  NA
## 23          2,773           100     93.04        35.79  NA  NA  NA
## 24          1,130           100     93.03        35.79  NA  NA  NA
## 25          3,362           100     92.47        35.68  NA  NA  NA
## 26          2,382           100     93.15        35.81  NA  NA  NA
## 27          2,207           100     92.84        35.75  NA  NA  NA
## 28            901           100     92.49        35.69  NA  NA  NA
## 29              0           100     89.11        34.91  NA  NA  NA
## 30          1,474           100     92.63        35.71  NA  NA  NA
## 31          1,467           100     92.64        35.72  NA  NA  NA
## 32            324           100     92.66        35.72  NA  NA  NA
## 33          3,222           100     92.79        35.74  NA  NA  NA
## 34          3,202           100     92.01        35.58  NA  NA  NA
## 35          1,704           100     91.87        35.56  NA  NA  NA
## 36          3,268           100     92.72        35.73  NA  NA  NA
## 37          2,626           100     92.32        35.65  NA  NA  NA
## 38          1,033           100     92.94        35.76  NA  NA  NA
## 39          1,453           100     92.30        35.65  NA  NA  NA
## 40          1,786           100     92.92        35.77  NA  NA  NA
## 41          2,509           100     91.86        35.55  NA  NA  NA
## 42          1,066           100     93.08        35.79  NA  NA  NA
## 43              0           100     95.85        36.37  NA  NA  NA
## 44          2,189           100     92.28        35.63  NA  NA  NA
## 45          2,086           100     91.90        35.56  NA  NA  NA
## 46          1,352           100     91.42        35.46  NA  NA  NA
## 47          3,631           100     93.06        35.79  NA  NA  NA
## 48          3,247           100     92.77        35.74  NA  NA  NA
## 49          3,931           100     92.61        35.70  NA  NA  NA
## 50          1,008           100     93.09        35.79  NA  NA  NA
## 51          2,062           100     92.67        35.71  NA  NA  NA
## 52          1,379           100     92.13        35.61  NA  NA  NA

ssCapAg <- ssCapAg[ssCapAg$Sample %in% colnames(x),]

dim(ssCapAg)

## [1] 48 31

x1 <- x[,which(colnames(x) %in% ssCapAg$Sample)]
dim(x1)

## [1] 310  48

x1 <- x1[which(rowMeans(x1)>10),]
dim(x1)

## [1] 261  48

ssCapAg %>%  kbl(caption="Sample sheet")%>% kable_styling("hover",full_width=FALSE) 

Sample sheet

	Lane	Project	Client.ID	Sample	X	Category	Sex.Age.Season	Sex	Age	Season	Location	Agebyyear	Weight	X.1	Number.of.Tumours	Ulcerated	UlcerationDepth	Secondary.infection	Total.tumour.volume	Barcode.sequence	PF.Clusters	X..of.the.lane	X..Perfect.barcode	X..One.mismatch.barcode	Yield..Mbases.	X..PF.Cluster	X…..Q30	Mean.Quality	X.2	X.3	X.4
1	1	default	S2_B99008	6102468	NA	Captive	FAU	F	A	Summer	Monarto	3.7369863	8.12	NA	NA	NA		NA	NA	GCTATCCT+AACAGGTG	5,338,890	1.00	98.08	1.92	1,612	100	93.69	35.91	NA	NA	NA
2	1	default	S2_Bonney	6102469	NA	Captive	FAU	F	A	Summer	Healesville	2.5863014	5.4	NA	NA	NA		NA	NA	ACGTTCAG+AGTTGTGC	14,968,459	3.00	97.82	2.18	4,520	100	93.68	35.91	NA	NA	NA
3	1	default	S3_Jessie	6102476	NA	Captive	FAW	F	A	Winter	Monarto	2.1945205	12.95	NA	NA	NA		NA	NA	CCGTAAGA+AGAGTCCA	4,340,498	1.00	97.95	2.05	1,311	100	93.54	35.88	NA	NA	NA
4	1	default	S3_McCauley	6102477	NA	Captive	MAA	M	A	Autumn	Healesville	3.0328767	8.52	NA	NA	NA		NA	NA	CAACACCT+AGCTACCA	16,944,152	3.00	97.98	2.02	5,117	100	93.05	35.78	NA	NA	NA
5	1	default	S5_Fenton	6102484	NA	Captive	MAU	M	A	Summer	Monarto	3.0000000	N/A	NA	NA	NA		NA	NA	GATTACCG+TTGGACTG	5,125,566	1.00	96.21	3.79	1,548	100	93.89	35.94	NA	NA	NA
6	1	default	S5_Bryan Adams	6102485	NA	Captive	MAW	M	A	Winter	Healesville	4.1616438	8.6	NA	NA	NA		NA	NA	CACTAGCT+TCGGATTC	5,317,767	1.00	97.33	2.67	1,606	100	93.24	35.83	NA	NA	NA
7	1	default	S7_Fredo	6102496	NA	Captive	MAA	M	A	Autumn	Healesville	2.9945205	8.8	NA	NA	NA		NA	NA	AGTGGATC+TCGATGAC	2,690,632	1.00	97.77	2.23	813	100	93.77	35.93	NA	NA	NA
8	1	default	S7_Piglet	6102497	NA	Captive	FSU	F	S	Summer	Healesville	1.5890411	4.76	NA	NA	NA		NA	NA	ATCACACG+GTAGCGTA	12,134,750	2.00	96.94	3.06	3,665	100	93.34	35.84	NA	NA	NA
9	1	default	S10_McMahon	6102514	NA	Captive	MSP	M	S	Spring	Healesville	1.6493151	8.06	NA	NA	NA		NA	NA	AACGTCTG+AATGACGC	3,809,607	1.00	96.73	3.27	1,151	100	93.07	35.79	NA	NA	NA
10	1	default	S10_Logan	6102515	NA	Captive	MAU	M	A	Summer	Healesville	5.6000000	9.02	NA	NA	NA		NA	NA	GTGCCATA+CTCCTAGT	9,606,720	2.00	97.52	2.48	2,901	100	93.25	35.83	NA	NA	NA
11	1	default	S11_Xuki	6102520	NA	Captive	MAA	M	A	Autumn	Monarto	2.0767123	7.95	NA	NA	NA		NA	NA	ATCGCCAT+TGCGTAAC	6,018,884	1.00	97.88	2.12	1,818	100	93.58	35.89	NA	NA	NA
12	1	default	S11_Kalamata	6102521	NA	Captive	MSW	M	S	Winter	Healesville	1.2849315	5.58	NA	NA	NA		NA	NA	GAACATCG+TCGACAAG	6,861,544	1.00	97.76	2.24	2,072	100	93.49	35.87	NA	NA	NA
13	1	default	S12_Wally	6102528	NA	Captive	MAA	M	A	Autumn	Monarto	2.1616438	N/A	NA	NA	NA		NA	NA	ACGACTTG+AACAGCGA	3,000,501	1.00	97.96	2.04	906	100	93.70	35.91	NA	NA	NA
14	1	default	S12_Naz	6102529	NA	Captive	MAW	M	A	Winter	Monarto	4.4547945	N/A	NA	NA	NA		NA	NA	CCTTGTAG+CCTTGGAA	5,927,942	1.00	97.40	2.60	1,790	100	93.86	35.94	NA	NA	NA
15	1	default	S13_Humbug	6102530	NA	Captive	MAU	M	A	Summer	Monarto	2.5397260	7.94	NA	NA	NA		NA	NA	GAGCAGTA+TGAGCTGT	1,702,608	0.00	98.07	1.93	514	100	93.58	35.89	NA	NA	NA
16	1	default	S13_Xerxes	6102535	NA	Captive	MAA	M	A	Autumn	Monarto	2.0767123	N/A	NA	NA	NA		NA	NA	GATAGGCT+CAGTCACA	7,628,257	2.00	98.34	1.66	2,304	100	93.62	35.89	NA	NA	NA
17	1	default	S14_Catani	6102538	NA	Captive	FAU	F	A	Summer	Healesville	2.8630137	6.46	NA	NA	NA		NA	NA	TGAAGACG+ACATGCCA	13,381,058	3.00	96.98	3.02	4,041	100	93.36	35.84	NA	NA	NA
18	2	default	S15_Amelia	6102541	NA	Captive	FAU	F	A	Summer	Healesville	4.7260274	5.85	NA	NA	NA		NA	NA	AACGTCTG+AATGACGC	8,041,818	1.73	96.32	3.68	2,429	100	92.18	35.63	NA	NA	NA
19	2	default	S15_London	6102542	NA	Captive	FAU	F	A	Summer	Healesville	3.7863014	4.8	NA	NA	NA		NA	NA	GTCATCGA+GATCCACT	5,758,111	1.24	96.72	3.28	1,739	100	91.87	35.55	NA	NA	NA
20	2	default	S15_Elyse	6102543	NA	Captive	FAP	F	A	Spring	Healesville	4.5232877	7.88	NA	NA	NA		NA	NA	AGCGTGTT+CTTCGCAA	5,774,637	1.24	96.76	3.24	1,744	100	91.56	35.50	NA	NA	NA
22	2	default	S16_Milla	6102548	NA	Captive	FAU	F	A	Summer	Healesville	2.7671233	5.82	NA	NA	NA		NA	NA	TTGCGAAG+AACACGCT	6,376,997	1.37	97.60	2.40	1,926	100	92.44	35.67	NA	NA	NA
23	2	default	S16_Maleficent	6102549	NA	Captive	FAU	F	A	Summer	Monarto	4.5589041	8.59	NA	NA	NA		NA	NA	TGCTTCCA+CGATCGAT	9,181,668	1.98	98.08	1.92	2,773	100	93.04	35.79	NA	NA	NA
24	2	default	S17_Anirak	6102553	NA	Captive	FSW	F	S	Winter	Healesville	1.3424658	5.42	NA	NA	NA		NA	NA	TGAGCTAG+GAACGGTT	3,740,878	0.81	97.15	2.85	1,130	100	93.03	35.79	NA	NA	NA
25	2	default	S17_Poly	6102554	NA	Captive	FSU	F	S	Summer	Monarto	1.7917808	6.67	NA	NA	NA		NA	NA	CTTAGGAC+ACTCCTAC	11,133,566	2.40	97.85	2.15	3,362	100	92.47	35.68	NA	NA	NA
26	2	default	S17_Methyl	6102556	NA	Captive	FSU	F	S	Summer	Healesville	0.7534247	2.72	NA	NA	NA		NA	NA	TGGTACAG+TCGAGAGT	7,887,375	1.70	97.57	2.43	2,382	100	93.15	35.81	NA	NA	NA
27	2	default	S18_Zara	6102558	NA	Captive	FSW	F	S	Winter	Monarto	1.5150685	4.6	NA	NA	NA		NA	NA	ACCACGAT+GTCTGCAA	7,308,907	1.58	97.70	2.30	2,207	100	92.84	35.75	NA	NA	NA
28	2	default	S18_Maze	6102559	NA	Captive	FSW	F	S	Winter	Healesville	1.4000000	5.82	NA	NA	NA		NA	NA	ACTGAGGT+GTCAACAG	2,984,930	0.64	97.59	2.41	901	100	92.49	35.69	NA	NA	NA
30	2	default	S19_Morocco	6102564	NA	Captive	MAU	M	A	Summer	Monarto	4.5917808	9.82	NA	NA	NA		NA	NA	TCAGGCTT+ATCATGCG	4,879,629	1.05	97.38	2.62	1,474	100	92.63	35.71	NA	NA	NA
31	2	default	S19_Pecorino	6102565	NA	Captive	FSW	F	S	Winter	Healesville	1.1123288	6.02	NA	NA	NA		NA	NA	ATGCCTGT+AGATTGCG	4,857,395	1.05	97.58	2.42	1,467	100	92.64	35.72	NA	NA	NA
33	2	default	S20_Muasy	6102569	NA	Captive	MSU	M	S	Summer	Healesville	1.4219178	6.62	NA	NA	NA		NA	NA	TTCCAAGG+CTACAAGG	10,669,237	2.30	97.64	2.36	3,222	100	92.79	35.74	NA	NA	NA
34	2	default	S20_Fuse	6102570	NA	Captive	MSU	M	S	Summer	Healesville	1.7315068	6.26	NA	NA	NA		NA	NA	GTAGGAGT+GTCCTAAG	10,603,179	2.29	97.34	2.66	3,202	100	92.01	35.58	NA	NA	NA
35	2	default	S20_Joseph	6102571	NA	Captive	MSU	M	S	Summer	Healesville	1.7479452	7.42	NA	NA	NA		NA	NA	CTTCGACT+GATCTTGC	5,642,025	1.22	96.37	3.63	1,704	100	91.87	35.56	NA	NA	NA
36	2	default	S21_Moama	6102576	NA	Captive	MAW	M	A	Winter	Monarto	1.6986301	8.75	NA	NA	NA		NA	NA	TTGGACGT+GAAGTGCT	10,821,895	2.33	96.02	3.98	3,268	100	92.72	35.73	NA	NA	NA
37	2	default	S21_Nauset	6102577	NA	Captive	MSW	M	S	Winter	Healesville	1.3041096	4.036	NA	NA	NA		NA	NA	TATCGGTC+TACCGGAT	8,693,920	1.87	98.13	1.87	2,626	100	92.32	35.65	NA	NA	NA
38	2	default	S21_Graham	6102578	NA	Captive	MSW	M	S	Winter	Healesville	1.3616438	6.4	NA	NA	NA		NA	NA	TACGCCTT+GCTACTCT	3,422,010	0.74	97.72	2.28	1,033	100	92.94	35.76	NA	NA	NA
39	2	default	S22_Yanni	6102581	NA	Captive	MSW	M	S	Winter	Monarto	1.5178082	8.15	NA	NA	NA		NA	NA	CGTGTGTA+CCAACGAA	4,811,453	1.04	97.67	2.33	1,453	100	92.30	35.65	NA	NA	NA
40	2	default	S22_Gherkin	6102582	NA	Captive	MSW	M	S	Winter	Healesville	1.2630137	6.32	NA	NA	NA		NA	NA	CCTTGTAG+CCTTGGAA	5,913,586	1.27	97.05	2.95	1,786	100	92.92	35.77	NA	NA	NA
41	2	default	S22_Stilton	6102584	NA	Captive	MSW	M	S	Winter	Healesville	1.2986301	5.76	NA	NA	NA		NA	NA	CGCAATCT+ACAGGCAT	8,306,537	1.79	96.94	3.06	2,509	100	91.86	35.55	NA	NA	NA
42	2	default	S23_Osiris	6102588	NA	Captive	MSW	M	S	Winter	Healesville	1.1753425	5.34	NA	NA	NA		NA	NA	AGTGGATC+TCGATGAC	3,528,471	0.76	97.24	2.76	1,066	100	93.08	35.79	NA	NA	NA
44	2	default	S23_Quince	6102590	NA	Captive	MSW	M	S	Winter	Healesville	1.3835616	N/A	NA	NA	NA		NA	NA	AACCGTTC+CTAGCTCA	7,247,793	1.56	97.63	2.37	2,189	100	92.28	35.63	NA	NA	NA
45	2	default	S24_Malfoy	6102595	NA	Captive	MAW	M	A	Winter	Healesville	5.1041096	8.02	NA	NA	NA		NA	NA	TAACCGGT+ATCGTCTC	6,908,416	1.49	97.09	2.91	2,086	100	91.90	35.56	NA	NA	NA
46	2	default	S24_Licorice	6102596	NA	Captive	FSA	F	S	Autumn	Healesville	1.8246575	5.94	NA	NA	NA		NA	NA	TTGCAGAC+ATCGTGGT	4,475,461	0.96	95.05	4.95	1,352	100	91.42	35.46	NA	NA	NA
47	2	default	S25_Tofu	6102599	NA	Captive	FAU	F	A	Summer	Healesville	2.7452055	N/A	NA	NA	NA		NA	NA	GAGACGAT+ACCGGTTA	12,024,421	2.59	95.37	4.63	3,631	100	93.06	35.79	NA	NA	NA
48	2	default	S25_Colette	6102600	NA	Captive	FAP	F	A	Spring	Healesville	6.6273973	7.52	NA	NA	NA		NA	NA	CTTGTCGA+CGATGTTC	10,751,797	2.32	97.49	2.51	3,247	100	92.77	35.74	NA	NA	NA
49	2	default	S25_CannonBall	6102601	NA	Captive	MSW	M	S	Winter	Healesville	1.1479452	6.54	NA	NA	NA		NA	NA	TACGCTAC+CGTGTGAT	13,015,081	2.81	97.36	2.64	3,931	100	92.61	35.70	NA	NA	NA
50	2	default	S25_Cornelius	6102602	NA	Captive	MAU	M	A	Summer	Healesville	3.6904110	7.74	NA	NA	NA		NA	NA	GAATCCGA+AGCTAGTG	3,336,983	0.72	97.10	2.90	1,008	100	93.09	35.79	NA	NA	NA
51	2	default	S25_Mardook	6102603	NA	Captive	MAW	M	A	Winter	Healesville	4.3013699	9.16	NA	NA	NA		NA	NA	GCACAACT+CTGAACGT	6,826,313	1.47	97.66	2.34	2,062	100	92.67	35.71	NA	NA	NA
52	2	default	S25_Negroni	6102604	NA	Captive	FSP	F	S	Spring	Healesville	1.6767123	5.86	NA	NA	NA		NA	NA	TGTGACTG+AGCCTATC	4,566,180	0.98	97.68	2.32	1,379	100	92.13	35.61	NA	NA	NA

col <- as.character(as.numeric(grepl("A",ssCapAg$Age)))
col <- gsub("0","orange",col)
col <- gsub("1","brown",col)

mds<-plotMDS(x1,pch=19,cex=3,col=col,main="MDS plot: Comparison of Sub-Adult vs Adult Captive devils")
mtext("Sub-Adult=Orange, Adult=Brown")

ssCapAg$CapLoc <- grepl("Monarto",ssCapAg$Location)
ssCapAg$MvF <- factor(ssCapAg$Sex)
ssCapAg$SubAdvAd <- as.numeric(factor(ssCapAg$Age,levels=c("S","A"),ordered = TRUE))
ssCapAg$Season <- factor(ssCapAg$Season)
str(ssCapAg)

## 'data.frame':    48 obs. of  34 variables:
##  $ Lane                   : int  1 1 1 1 1 1 1 1 1 1 ...
##  $ Project                : chr  "default" "default" "default" "default" ...
##  $ Client.ID              : chr  "S2_B99008" "S2_Bonney" "S3_Jessie" "S3_McCauley" ...
##  $ Sample                 : int  6102468 6102469 6102476 6102477 6102484 6102485 6102496 6102497 6102514 6102515 ...
##  $ X                      : logi  NA NA NA NA NA NA ...
##  $ Category               : chr  "Captive" "Captive" "Captive" "Captive" ...
##  $ Sex.Age.Season         : chr  "FAU" "FAU" "FAW" "MAA" ...
##  $ Sex                    : chr  "F" "F" "F" "M" ...
##  $ Age                    : chr  "A" "A" "A" "A" ...
##  $ Season                 : Factor w/ 4 levels "Autumn","Spring",..: 3 3 4 1 3 4 1 3 2 3 ...
##  $ Location               : chr  "Monarto" "Healesville" "Monarto" "Healesville" ...
##  $ Agebyyear              : num  3.74 2.59 2.19 3.03 3 ...
##  $ Weight                 : chr  "8.12" "5.4" "12.95" "8.52" ...
##  $ X.1                    : logi  NA NA NA NA NA NA ...
##  $ Number.of.Tumours      : int  NA NA NA NA NA NA NA NA NA NA ...
##  $ Ulcerated              : logi  NA NA NA NA NA NA ...
##  $ UlcerationDepth        : chr  "" "" "" "" ...
##  $ Secondary.infection    : logi  NA NA NA NA NA NA ...
##  $ Total.tumour.volume    : num  NA NA NA NA NA NA NA NA NA NA ...
##  $ Barcode.sequence       : chr  "GCTATCCT+AACAGGTG" "ACGTTCAG+AGTTGTGC" "CCGTAAGA+AGAGTCCA" "CAACACCT+AGCTACCA" ...
##  $ PF.Clusters            : chr  "5,338,890" "14,968,459" "4,340,498" "16,944,152" ...
##  $ X..of.the.lane         : num  1 3 1 3 1 1 1 2 1 2 ...
##  $ X..Perfect.barcode     : num  98.1 97.8 98 98 96.2 ...
##  $ X..One.mismatch.barcode: num  1.92 2.18 2.05 2.02 3.79 2.67 2.23 3.06 3.27 2.48 ...
##  $ Yield..Mbases.         : chr  "1,612" "4,520" "1,311" "5,117" ...
##  $ X..PF.Cluster          : int  100 100 100 100 100 100 100 100 100 100 ...
##  $ X.....Q30              : num  93.7 93.7 93.5 93 93.9 ...
##  $ Mean.Quality           : num  35.9 35.9 35.9 35.8 35.9 ...
##  $ X.2                    : logi  NA NA NA NA NA NA ...
##  $ X.3                    : logi  NA NA NA NA NA NA ...
##  $ X.4                    : logi  NA NA NA NA NA NA ...
##  $ CapLoc                 : logi  TRUE FALSE TRUE FALSE TRUE FALSE ...
##  $ MvF                    : Factor w/ 2 levels "F","M": 1 1 1 2 2 2 2 1 2 2 ...
##  $ SubAdvAd               : num  2 2 2 2 2 2 2 1 1 2 ...

dds <- DESeqDataSetFromMatrix(countData = x1 , colData = ssCapAg, design = ~ Season + MvF + CapLoc + SubAdvAd)

## converting counts to integer mode

##   the design formula contains one or more numeric variables with integer values,
##   specifying a model with increasing fold change for higher values.
##   did you mean for this to be a factor? if so, first convert
##   this variable to a factor using the factor() function

res <- DESeq(dds)

## estimating size factors

## estimating dispersions

## gene-wise dispersion estimates

## mean-dispersion relationship

## final dispersion estimates

## fitting model and testing

## -- replacing outliers and refitting for 10 genes
## -- DESeq argument 'minReplicatesForReplace' = 7 
## -- original counts are preserved in counts(dds)

## estimating dispersions

## fitting model and testing

z<- results(res)
vsd <- vst(dds, blind=FALSE,nsub=nrow(x1)-50)
zz<-cbind(as.data.frame(z),assay(vsd))
dge<-as.data.frame(zz[order(zz$pvalue),])
dgeCapAg <- dge
sig <- subset(dge,padj<0.05)
sig1_up <- rownames(subset(sig,log2FoldChange>0))
sig1_dn <- rownames(subset(sig,log2FoldChange<0))
length(sig1_up)

## [1] 14

maplot(dgeCapAg,"Sub-Adult vs Adult Captive devils")

make_volcano(dgeCapAg,"Sub-Adult vs Adult Captive devils")

col <- as.character(as.numeric(grepl("A",ssCapAg$Age)))
col <- gsub("0","orange",col)
col <- gsub("1","brown",col)
mtext("Sub-Adult=Orange, Adult=Brown")

sig[1:100,1:6] %>%
  kbl(caption="Comparison of Sub-Adult vs Adult Captive devils") %>%
  kable_paper("hover", full_width = F)

Comparison of Sub-Adult vs Adult Captive devils

	baseMean	log2FoldChange	lfcSE	stat	pvalue	padj
FLT3	2665.40411	-1.2796830	0.2642011	-4.843594	0.0000013	0.0003264
IL5	660.81847	-3.4300117	0.7715965	-4.445344	0.0000088	0.0010884
GATA2	1727.77825	1.0974081	0.2514531	4.364266	0.0000128	0.0010884
CIITA	3093.73316	-0.7904557	0.1891149	-4.179765	0.0000292	0.0015356
Saha-UC	440247.70240	1.0594366	0.2548969	4.156334	0.0000323	0.0015356
IgA	31617.57508	0.9692246	0.2345756	4.131822	0.0000360	0.0015356
Saha-UD	766155.33670	1.5843532	0.3875401	4.088231	0.0000435	0.0015802
CCL24	1456.99179	1.0838345	0.2672030	4.056222	0.0000499	0.0015802
Saha-UA	295824.31182	0.9194355	0.2280948	4.030937	0.0000556	0.0015802
IRF8	3207.91842	-0.6195137	0.1618325	-3.828118	0.0001291	0.0031680
CD83	201.68746	-1.0292791	0.2697905	-3.815104	0.0001361	0.0031680
CCL25	24.47158	-3.1574774	0.8341792	-3.785131	0.0001536	0.0032774
IgM	151322.22005	-0.8119283	0.2162946	-3.753807	0.0001742	0.0033610
TLR7	1413.02239	-0.6398268	0.1710633	-3.740293	0.0001838	0.0033610
TLR3	1013.97122	-0.5247096	0.1434902	-3.656763	0.0002554	0.0043166
IRF5	6518.67692	-0.6716279	0.1843762	-3.642704	0.0002698	0.0043166
CD22	21934.33484	-0.7279433	0.2056778	-3.539241	0.0004013	0.0060208
STING1	1617.54127	-0.6813284	0.1932797	-3.525091	0.0004233	0.0060208
F10	170051.52401	1.2010900	0.3512393	3.419577	0.0006272	0.0084505
CD19	6708.07000	-0.7973806	0.2382073	-3.347423	0.0008157	0.0095777
B2M	265069.64285	1.0050865	0.3004285	3.345509	0.0008213	0.0095777
CCR1	527.97713	0.8076812	0.2414656	3.344912	0.0008231	0.0095777
PDCD1	2203.24533	-0.7944444	0.2390848	-3.322856	0.0008910	0.0099173
Rag1	46.82983	-2.2895610	0.6987534	-3.276637	0.0010505	0.0112055
Saha-UB	345632.06581	0.8722311	0.2731084	3.193717	0.0014045	0.0143825
CD40	5525.72470	-0.7288007	0.2308827	-3.156584	0.0015963	0.0155993
IL21R	7824.98893	-0.6433001	0.2046134	-3.143978	0.0016667	0.0155993
CCL20	33.90915	-3.1671242	1.0101290	-3.135366	0.0017164	0.0155993
Saha-DAA	140702.30852	-0.4803360	0.1536183	-3.126815	0.0017671	0.0155993
OAZ1	391338.51842	1.1332851	0.3678947	3.080460	0.0020668	0.0176368
FCMR	67644.98037	-0.8637685	0.2815813	-3.067563	0.0021581	0.0178219
C5	119.45516	-1.0963492	0.3656139	-2.998653	0.0027118	0.0216941
Saha-DAB2	18376.93729	-0.5553659	0.1894875	-2.930884	0.0033800	0.0258483
CD36	12970.68540	0.9212163	0.3148330	2.926048	0.0034330	0.0258483
FCER1A	351.01967	1.0182569	0.3519379	2.893286	0.0038123	0.0278845
IL24	40.34769	-1.8196785	0.6453297	-2.819766	0.0048059	0.0341751
EIF2AK2	855.79939	-0.5424791	0.1939458	-2.797066	0.0051569	0.0356802
MS4A1	15233.57533	-0.6784880	0.2440810	-2.779766	0.0054398	0.0366471
IL22	29.18288	-2.8152596	1.0359161	-2.717652	0.0065747	0.0431569
FGF2	127.33242	-0.6207586	0.2295239	-2.704549	0.0068397	0.0437742
IL10RA	22527.62924	-0.5408429	0.2007779	-2.693737	0.0070656	0.0441169
CD9	59462.73116	1.0485088	0.3929423	2.668353	0.0076224	0.0464604
NA	NA	NA	NA	NA	NA	NA
NA.1	NA	NA	NA	NA	NA	NA
NA.2	NA	NA	NA	NA	NA	NA
NA.3	NA	NA	NA	NA	NA	NA
NA.4	NA	NA	NA	NA	NA	NA
NA.5	NA	NA	NA	NA	NA	NA
NA.6	NA	NA	NA	NA	NA	NA
NA.7	NA	NA	NA	NA	NA	NA
NA.8	NA	NA	NA	NA	NA	NA
NA.9	NA	NA	NA	NA	NA	NA
NA.10	NA	NA	NA	NA	NA	NA
NA.11	NA	NA	NA	NA	NA	NA
NA.12	NA	NA	NA	NA	NA	NA
NA.13	NA	NA	NA	NA	NA	NA
NA.14	NA	NA	NA	NA	NA	NA
NA.15	NA	NA	NA	NA	NA	NA
NA.16	NA	NA	NA	NA	NA	NA
NA.17	NA	NA	NA	NA	NA	NA
NA.18	NA	NA	NA	NA	NA	NA
NA.19	NA	NA	NA	NA	NA	NA
NA.20	NA	NA	NA	NA	NA	NA
NA.21	NA	NA	NA	NA	NA	NA
NA.22	NA	NA	NA	NA	NA	NA
NA.23	NA	NA	NA	NA	NA	NA
NA.24	NA	NA	NA	NA	NA	NA
NA.25	NA	NA	NA	NA	NA	NA
NA.26	NA	NA	NA	NA	NA	NA
NA.27	NA	NA	NA	NA	NA	NA
NA.28	NA	NA	NA	NA	NA	NA
NA.29	NA	NA	NA	NA	NA	NA
NA.30	NA	NA	NA	NA	NA	NA
NA.31	NA	NA	NA	NA	NA	NA
NA.32	NA	NA	NA	NA	NA	NA
NA.33	NA	NA	NA	NA	NA	NA
NA.34	NA	NA	NA	NA	NA	NA
NA.35	NA	NA	NA	NA	NA	NA
NA.36	NA	NA	NA	NA	NA	NA
NA.37	NA	NA	NA	NA	NA	NA
NA.38	NA	NA	NA	NA	NA	NA
NA.39	NA	NA	NA	NA	NA	NA
NA.40	NA	NA	NA	NA	NA	NA
NA.41	NA	NA	NA	NA	NA	NA
NA.42	NA	NA	NA	NA	NA	NA
NA.43	NA	NA	NA	NA	NA	NA
NA.44	NA	NA	NA	NA	NA	NA
NA.45	NA	NA	NA	NA	NA	NA
NA.46	NA	NA	NA	NA	NA	NA
NA.47	NA	NA	NA	NA	NA	NA
NA.48	NA	NA	NA	NA	NA	NA
NA.49	NA	NA	NA	NA	NA	NA
NA.50	NA	NA	NA	NA	NA	NA
NA.51	NA	NA	NA	NA	NA	NA
NA.52	NA	NA	NA	NA	NA	NA
NA.53	NA	NA	NA	NA	NA	NA
NA.54	NA	NA	NA	NA	NA	NA
NA.55	NA	NA	NA	NA	NA	NA
NA.56	NA	NA	NA	NA	NA	NA
NA.57	NA	NA	NA	NA	NA	NA

write.table(dge,file="dgeCapAg.tsv",sep="\t")

mx <- sig[,7:ncol(sig)]
mx <- head(mx,30)
colfunc <- colorRampPalette(c("blue", "white", "red"))
heatmap.2(as.matrix(mx),trace="none",scale="row",
          col=colfunc(25),ColSideColors=col,mar=c(5,12)) 

ssH <- filter(ss, Category %in% c('Healthy','Healthy/Wild'))

ssHS <- ssH[which(grepl("M",ssH$Sex) + grepl("F",ssH$Sex)>0),]
ssHS

##    Lane Project     Client.ID  Sample  X     Category Sex.Age.Season Sex Age
## 1     1 default S1_LauraGrace 6102461 NA Healthy/Wild            FAP   F   A
## 2     1 default   S1_Turmeric 6102465 NA Healthy/Wild            FAU   F   A
## 3     1 default  S1_Singapore 6102466 NA      Healthy            FAA   F   A
## 4     1 default    S2_Belinda 6102467 NA Healthy/Wild            FAU   F   A
## 5     1 default      S2_Rosie 6102470 NA Healthy/Wild            FAU   F   A
## 6     1 default       S3_ST62 6102475 NA Healthy/Wild            FAW   F   A
## 7     1 default    S4_Kerchak 6102478 NA      Healthy            MAA   M   A
## 8     1 default    S4_Duffman 6102479 NA      Healthy            MAA   M   A
## 9     1 default      S4_Hogan 6102481 NA      Healthy            MAA   M   A
## 10    1 default      S5_Silky 6102486 NA Healthy/Wild            MAW   M   A
## 11    1 default    S5_Longley 6102487 NA Healthy/Wild            MAU   M   A
## 12    1 default    S6_Myanmar 6102490 NA      Healthy            FAP   F   A
## 13    1 default     S6_Pakana 6102491 NA      Healthy            MSP   M   S
## 14    1 default     S6_Hobbit 6102495 NA      Healthy            MSP   M   S
## 15    1 default      S7_Lieke 6102499 NA Healthy/Wild            FSU   F   S
## 16    1 default   S7_Smithers 6102501 NA Healthy/Wild            MAA   M   A
## 17    1 default     S8_Vinnie 6102503 NA Healthy/Wild            MAU   M   A
## 18    1 default  S8_Struggles 6102504 NA Healthy/Wild            MAW   M   A
## 19    1 default     S8_SusieQ 6102506 NA Healthy/Wild            FAU   F   A
## 20    1 default     S9_Majula 6102509 NA      Healthy            FAA   F   A
## 21    1 default     S9_Snoopy 6102510 NA Healthy/Wild            MSW   M   S
## 22    1 default     S9_Sarabi 6102513 NA Healthy/Wild            FSP   F   S
## 23    1 default     S11_Baloo 6102518 NA Healthy/Wild            MAA   M   A
## 24    1 default S11_DanRhodes 6102523 NA Healthy/Wild            MSW   M   S
## 25    1 default     S12_Robin 6102525 NA Healthy/Wild            MAA   M   A
## 26    1 default    S12_Eeyore 6102527 NA Healthy/Wild            MAW   M   A
## 27    1 default    S13_Pangea 6102533 NA Healthy/Wild            MAU   M   A
## 28    1 default    S13_Finbar 6102534 NA Healthy/Wild            MAA   M   A
## 29    2 default    S22_Ginger 6102586 NA Healthy/Wild            MSW   M   S
## 30    2 default       S24_Boy 6102593 NA Healthy/Wild            MAA   M   A
## 31    2 default    S24_Shazza 6102598 NA      Healthy            FAW   F   A
##    Season Location Agebyyear Weight X.1 Number.of.Tumours Ulcerated
## 1  Spring  Channel 2.6219178    7.2  NA                NA        NA
## 2  Summer  Channel 1.9095890    7.8  NA                NA        NA
## 3  Autumn  Channel 5.0000000    8.6  NA                NA        NA
## 4  Summer  Channel 1.8657534    6.6  NA                NA        NA
## 5  Summer  Channel 1.8904110    7.4  NA                NA        NA
## 6  Winter  Channel 3.2712329   10.2  NA                NA        NA
## 7  Autumn  Channel 4.1863014   11.6  NA                NA        NA
## 8  Autumn  Channel 3.1123288   12.4  NA                NA        NA
## 9  Autumn  Channel 3.1616438   11.6  NA                NA        NA
## 10 Winter  Channel 2.2849315    9.5  NA                NA        NA
## 11 Summer  Channel 2.9315068    9.7  NA                NA        NA
## 12 Spring  Channel 5.5726027    7.6  NA                NA        NA
## 13 Spring  Channel 1.6712329    7.8  NA                NA        NA
## 14 Spring  Channel 1.6273973      9  NA                NA        NA
## 15 Summer  Channel 0.8657534    4.9  NA                NA        NA
## 16 Autumn  Channel 3.1013699    8.9  NA                NA        NA
## 17 Summer  Channel 2.9150685    8.8  NA                NA        NA
## 18 Winter  Channel 3.2821918   11.1  NA                NA        NA
## 19 Summer  Channel 1.8712329    7.1  NA                NA        NA
## 20 Autumn  Channel 3.1150685      6  NA                NA        NA
## 21 Winter  Channel 1.2794521    6.3  NA                NA        NA
## 22 Spring  Channel 1.6328767    5.8  NA                NA        NA
## 23 Autumn  Channel 3.1726027   11.5  NA                NA        NA
## 24 Winter  Channel 1.3917808    6.8  NA                NA        NA
## 25 Autumn  Channel 2.1972603    7.7  NA                NA        NA
## 26 Winter  Channel 3.2794521      9  NA                NA        NA
## 27 Summer  Channel 1.8794521    7.7  NA                NA        NA
## 28 Autumn  Channel 2.1150685    6.8  NA                NA        NA
## 29 Winter  Channel 1.2712329    7.6  NA                NA        NA
## 30 Autumn  Channel 3.1452055    8.3  NA                NA        NA
## 31 Winter  Channel 2.0273973    6.2  NA                NA        NA
##    UlcerationDepth Secondary.infection Total.tumour.volume  Barcode.sequence
## 1                                   NA                  NA AGAGTAGC+AAGGCGTA
## 2                                   NA                  NA GAAGGAAG+GACGTCAT
## 3                                   NA                  NA ACAGCTCA+TACTGCTC
## 4                                   NA                  NA GCAAGATC+AGTCGAAG
## 5                                   NA                  NA CGCATGAT+AAGCCTGA
## 6                                   NA                  NA TGAGCTAG+GAACGGTT
## 7                                   NA                  NA CGGCTAAT+CTCGTTCT
## 8                                   NA                  NA GACGATCT+TCGTGCAT
## 9                                   NA                  NA TACGCCTT+GCTACTCT
## 10                                  NA                  NA CTGTTGAC+ACCTCAGT
## 11                                  NA                  NA CGCAATCT+ACAGGCAT
## 12                                  NA                  NA GCACAACT+CTGAACGT
## 13                                  NA                  NA GTTACGCA+ATGGCGAT
## 14                                  NA                  NA ATATGCGC+ACGATCAG
## 15                                  NA                  NA TTCCAAGG+CTACAAGG
## 16                                  NA                  NA GCGTCATT+CAGACGTT
## 17                                  NA                  NA CGACGTTA+CTCTGGAT
## 18                                  NA                  NA ATCGATCG+TGGAAGCA
## 19                                  NA                  NA CAGTCCAA+CGGTAATC
## 20                                  NA                  NA ACTCTCGA+CTGTACCA
## 21                                  NA                  NA AACTGAGC+CAATCAGG
## 22                                  NA                  NA CTTAGGAC+ACTCCTAC
## 23                                  NA                  NA CTGATCGT+GCGCATAT
## 24                                  NA                  NA TCGCTGTT+CAAGTCGT
## 25                                  NA                  NA ATGACGTC+CTTCCTTC
## 26                                  NA                  NA TCAGGCTT+ATCATGCG
## 27                                  NA                  NA ACCACGAT+GTCTGCAA
## 28                                  NA                  NA GAGACGAT+ACCGGTTA
## 29                                  NA                  NA ACAGCTCA+TACTGCTC
## 30                                  NA                  NA CGACGTTA+CTCTGGAT
## 31                                  NA                  NA GAGCAGTA+TGAGCTGT
##    PF.Clusters X..of.the.lane X..Perfect.barcode X..One.mismatch.barcode
## 1    4,511,702           1.00              98.29                    1.71
## 2    6,479,047           1.00              97.28                    2.72
## 3    4,481,213           1.00              97.99                    2.01
## 4    3,296,547           1.00              97.40                    2.60
## 5    4,556,781           1.00              95.70                    4.30
## 6    4,579,681           1.00              96.67                    3.33
## 7    6,886,511           1.00              97.80                    2.20
## 8    8,301,336           2.00              97.75                    2.25
## 9    2,990,345           1.00              97.77                    2.23
## 10   5,617,260           1.00              97.81                    2.19
## 11   2,937,660           1.00              97.71                    2.29
## 12   9,604,175           2.00              98.14                    1.86
## 13   4,383,521           1.00              98.09                    1.91
## 14   7,342,414           2.00              97.00                    3.00
## 15   1,800,468           0.00              97.79                    2.21
## 16   6,731,177           1.00              98.04                    1.96
## 17   7,901,958           2.00              98.02                    1.98
## 18   5,257,820           1.00              95.13                    4.87
## 19     454,240           0.00              97.95                    2.05
## 20   8,490,270           2.00              97.99                    2.01
## 21   1,709,078           0.00              98.04                    1.96
## 22   3,945,235           1.00              98.11                    1.89
## 23  10,923,019           2.00              98.07                    1.93
## 24   5,076,673           1.00              97.41                    2.59
## 25   6,756,342           1.00              97.12                    2.88
## 26   3,071,283           1.00              97.28                    2.72
## 27   1,732,316           0.00              97.57                    2.43
## 28       1,152           0.00              96.70                    3.30
## 29   9,742,792           2.10              97.63                    2.37
## 30   1,476,427           0.32              97.63                    2.37
## 31   2,693,655           0.58              97.51                    2.49
##    Yield..Mbases. X..PF.Cluster X.....Q30 Mean.Quality X.2 X.3 X.4
## 1           1,363           100     93.82        35.92  NA  NA  NA
## 2           1,957           100     93.85        35.94  NA  NA  NA
## 3           1,353           100     93.95        35.96  NA  NA  NA
## 4             996           100     94.12        35.98  NA  NA  NA
## 5           1,376           100     93.94        35.94  NA  NA  NA
## 6           1,383           100     93.75        35.92  NA  NA  NA
## 7           2,080           100     93.61        35.89  NA  NA  NA
## 8           2,507           100     93.83        35.93  NA  NA  NA
## 9             903           100     93.71        35.90  NA  NA  NA
## 10          1,696           100     93.55        35.88  NA  NA  NA
## 11            887           100     93.75        35.91  NA  NA  NA
## 12          2,900           100     93.93        35.94  NA  NA  NA
## 13          1,324           100     93.90        35.95  NA  NA  NA
## 14          2,217           100     93.93        35.96  NA  NA  NA
## 15            544           100     93.58        35.88  NA  NA  NA
## 16          2,033           100     93.62        35.89  NA  NA  NA
## 17          2,386           100     94.09        35.98  NA  NA  NA
## 18          1,588           100     93.81        35.93  NA  NA  NA
## 19            137           100     92.97        35.76  NA  NA  NA
## 20          2,564           100     93.34        35.84  NA  NA  NA
## 21            516           100     93.06        35.78  NA  NA  NA
## 22          1,191           100     93.55        35.88  NA  NA  NA
## 23          3,299           100     93.51        35.87  NA  NA  NA
## 24          1,533           100     93.77        35.92  NA  NA  NA
## 25          2,040           100     92.53        35.68  NA  NA  NA
## 26            928           100     92.94        35.76  NA  NA  NA
## 27            523           100     93.82        35.92  NA  NA  NA
## 28              0           100     92.06        35.56  NA  NA  NA
## 29          2,942           100     93.00        35.78  NA  NA  NA
## 30            446           100     93.25        35.82  NA  NA  NA
## 31            813           100     93.01        35.79  NA  NA  NA

ssHS <- ssHS[ssHS$Sample %in% colnames(x),]

dim(ssHS)

## [1] 29 31

x1 <- x[,which(colnames(x) %in% ssHS$Sample)]
dim(x1)

## [1] 310  29

x1 <- x1[which(rowMeans(x1)>10),]
dim(x1)

## [1] 256  29

ssHS %>%  kbl(caption="Sample sheet")%>% kable_styling("hover",full_width=FALSE) 

Sample sheet

	Lane	Project	Client.ID	Sample	X	Category	Sex.Age.Season	Sex	Age	Season	Location	Agebyyear	Weight	X.1	Number.of.Tumours	Ulcerated	UlcerationDepth	Secondary.infection	Total.tumour.volume	Barcode.sequence	PF.Clusters	X..of.the.lane	X..Perfect.barcode	X..One.mismatch.barcode	Yield..Mbases.	X..PF.Cluster	X…..Q30	Mean.Quality	X.2	X.3	X.4
1	1	default	S1_LauraGrace	6102461	NA	Healthy/Wild	FAP	F	A	Spring	Channel	2.6219178	7.2	NA	NA	NA		NA	NA	AGAGTAGC+AAGGCGTA	4,511,702	1.00	98.29	1.71	1,363	100	93.82	35.92	NA	NA	NA
2	1	default	S1_Turmeric	6102465	NA	Healthy/Wild	FAU	F	A	Summer	Channel	1.9095890	7.8	NA	NA	NA		NA	NA	GAAGGAAG+GACGTCAT	6,479,047	1.00	97.28	2.72	1,957	100	93.85	35.94	NA	NA	NA
3	1	default	S1_Singapore	6102466	NA	Healthy	FAA	F	A	Autumn	Channel	5.0000000	8.6	NA	NA	NA		NA	NA	ACAGCTCA+TACTGCTC	4,481,213	1.00	97.99	2.01	1,353	100	93.95	35.96	NA	NA	NA
4	1	default	S2_Belinda	6102467	NA	Healthy/Wild	FAU	F	A	Summer	Channel	1.8657534	6.6	NA	NA	NA		NA	NA	GCAAGATC+AGTCGAAG	3,296,547	1.00	97.40	2.60	996	100	94.12	35.98	NA	NA	NA
5	1	default	S2_Rosie	6102470	NA	Healthy/Wild	FAU	F	A	Summer	Channel	1.8904110	7.4	NA	NA	NA		NA	NA	CGCATGAT+AAGCCTGA	4,556,781	1.00	95.70	4.30	1,376	100	93.94	35.94	NA	NA	NA
6	1	default	S3_ST62	6102475	NA	Healthy/Wild	FAW	F	A	Winter	Channel	3.2712329	10.2	NA	NA	NA		NA	NA	TGAGCTAG+GAACGGTT	4,579,681	1.00	96.67	3.33	1,383	100	93.75	35.92	NA	NA	NA
7	1	default	S4_Kerchak	6102478	NA	Healthy	MAA	M	A	Autumn	Channel	4.1863014	11.6	NA	NA	NA		NA	NA	CGGCTAAT+CTCGTTCT	6,886,511	1.00	97.80	2.20	2,080	100	93.61	35.89	NA	NA	NA
8	1	default	S4_Duffman	6102479	NA	Healthy	MAA	M	A	Autumn	Channel	3.1123288	12.4	NA	NA	NA		NA	NA	GACGATCT+TCGTGCAT	8,301,336	2.00	97.75	2.25	2,507	100	93.83	35.93	NA	NA	NA
9	1	default	S4_Hogan	6102481	NA	Healthy	MAA	M	A	Autumn	Channel	3.1616438	11.6	NA	NA	NA		NA	NA	TACGCCTT+GCTACTCT	2,990,345	1.00	97.77	2.23	903	100	93.71	35.90	NA	NA	NA
10	1	default	S5_Silky	6102486	NA	Healthy/Wild	MAW	M	A	Winter	Channel	2.2849315	9.5	NA	NA	NA		NA	NA	CTGTTGAC+ACCTCAGT	5,617,260	1.00	97.81	2.19	1,696	100	93.55	35.88	NA	NA	NA
11	1	default	S5_Longley	6102487	NA	Healthy/Wild	MAU	M	A	Summer	Channel	2.9315068	9.7	NA	NA	NA		NA	NA	CGCAATCT+ACAGGCAT	2,937,660	1.00	97.71	2.29	887	100	93.75	35.91	NA	NA	NA
12	1	default	S6_Myanmar	6102490	NA	Healthy	FAP	F	A	Spring	Channel	5.5726027	7.6	NA	NA	NA		NA	NA	GCACAACT+CTGAACGT	9,604,175	2.00	98.14	1.86	2,900	100	93.93	35.94	NA	NA	NA
13	1	default	S6_Pakana	6102491	NA	Healthy	MSP	M	S	Spring	Channel	1.6712329	7.8	NA	NA	NA		NA	NA	GTTACGCA+ATGGCGAT	4,383,521	1.00	98.09	1.91	1,324	100	93.90	35.95	NA	NA	NA
14	1	default	S6_Hobbit	6102495	NA	Healthy	MSP	M	S	Spring	Channel	1.6273973	9	NA	NA	NA		NA	NA	ATATGCGC+ACGATCAG	7,342,414	2.00	97.00	3.00	2,217	100	93.93	35.96	NA	NA	NA
15	1	default	S7_Lieke	6102499	NA	Healthy/Wild	FSU	F	S	Summer	Channel	0.8657534	4.9	NA	NA	NA		NA	NA	TTCCAAGG+CTACAAGG	1,800,468	0.00	97.79	2.21	544	100	93.58	35.88	NA	NA	NA
16	1	default	S7_Smithers	6102501	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	3.1013699	8.9	NA	NA	NA		NA	NA	GCGTCATT+CAGACGTT	6,731,177	1.00	98.04	1.96	2,033	100	93.62	35.89	NA	NA	NA
17	1	default	S8_Vinnie	6102503	NA	Healthy/Wild	MAU	M	A	Summer	Channel	2.9150685	8.8	NA	NA	NA		NA	NA	CGACGTTA+CTCTGGAT	7,901,958	2.00	98.02	1.98	2,386	100	94.09	35.98	NA	NA	NA
18	1	default	S8_Struggles	6102504	NA	Healthy/Wild	MAW	M	A	Winter	Channel	3.2821918	11.1	NA	NA	NA		NA	NA	ATCGATCG+TGGAAGCA	5,257,820	1.00	95.13	4.87	1,588	100	93.81	35.93	NA	NA	NA
20	1	default	S9_Majula	6102509	NA	Healthy	FAA	F	A	Autumn	Channel	3.1150685	6	NA	NA	NA		NA	NA	ACTCTCGA+CTGTACCA	8,490,270	2.00	97.99	2.01	2,564	100	93.34	35.84	NA	NA	NA
21	1	default	S9_Snoopy	6102510	NA	Healthy/Wild	MSW	M	S	Winter	Channel	1.2794521	6.3	NA	NA	NA		NA	NA	AACTGAGC+CAATCAGG	1,709,078	0.00	98.04	1.96	516	100	93.06	35.78	NA	NA	NA
22	1	default	S9_Sarabi	6102513	NA	Healthy/Wild	FSP	F	S	Spring	Channel	1.6328767	5.8	NA	NA	NA		NA	NA	CTTAGGAC+ACTCCTAC	3,945,235	1.00	98.11	1.89	1,191	100	93.55	35.88	NA	NA	NA
23	1	default	S11_Baloo	6102518	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	3.1726027	11.5	NA	NA	NA		NA	NA	CTGATCGT+GCGCATAT	10,923,019	2.00	98.07	1.93	3,299	100	93.51	35.87	NA	NA	NA
24	1	default	S11_DanRhodes	6102523	NA	Healthy/Wild	MSW	M	S	Winter	Channel	1.3917808	6.8	NA	NA	NA		NA	NA	TCGCTGTT+CAAGTCGT	5,076,673	1.00	97.41	2.59	1,533	100	93.77	35.92	NA	NA	NA
25	1	default	S12_Robin	6102525	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	2.1972603	7.7	NA	NA	NA		NA	NA	ATGACGTC+CTTCCTTC	6,756,342	1.00	97.12	2.88	2,040	100	92.53	35.68	NA	NA	NA
26	1	default	S12_Eeyore	6102527	NA	Healthy/Wild	MAW	M	A	Winter	Channel	3.2794521	9	NA	NA	NA		NA	NA	TCAGGCTT+ATCATGCG	3,071,283	1.00	97.28	2.72	928	100	92.94	35.76	NA	NA	NA
27	1	default	S13_Pangea	6102533	NA	Healthy/Wild	MAU	M	A	Summer	Channel	1.8794521	7.7	NA	NA	NA		NA	NA	ACCACGAT+GTCTGCAA	1,732,316	0.00	97.57	2.43	523	100	93.82	35.92	NA	NA	NA
29	2	default	S22_Ginger	6102586	NA	Healthy/Wild	MSW	M	S	Winter	Channel	1.2712329	7.6	NA	NA	NA		NA	NA	ACAGCTCA+TACTGCTC	9,742,792	2.10	97.63	2.37	2,942	100	93.00	35.78	NA	NA	NA
30	2	default	S24_Boy	6102593	NA	Healthy/Wild	MAA	M	A	Autumn	Channel	3.1452055	8.3	NA	NA	NA		NA	NA	CGACGTTA+CTCTGGAT	1,476,427	0.32	97.63	2.37	446	100	93.25	35.82	NA	NA	NA
31	2	default	S24_Shazza	6102598	NA	Healthy	FAW	F	A	Winter	Channel	2.0273973	6.2	NA	NA	NA		NA	NA	GAGCAGTA+TGAGCTGT	2,693,655	0.58	97.51	2.49	813	100	93.01	35.79	NA	NA	NA

col <- as.character(as.numeric(grepl("F",ssHS$Sex)))
col <- gsub("0","grey",col)
col <- gsub("1","black",col)

mds<-plotMDS(x1,pch=19,cex=3,col=col,main="MDS plot: Comparison of Male and Female Healthy devils")
mtext("Male=Grey, Female=Black")

ssHS$MvF <- factor(ssHS$Sex)
ssHS$SubAdvAd <- as.numeric(factor(ssHS$Age,levels=c("S","A"),ordered = TRUE))
ssHS$Season <- factor(ssHS$Season)
str(ssHS)

## 'data.frame':    29 obs. of  33 variables:
##  $ Lane                   : int  1 1 1 1 1 1 1 1 1 1 ...
##  $ Project                : chr  "default" "default" "default" "default" ...
##  $ Client.ID              : chr  "S1_LauraGrace" "S1_Turmeric" "S1_Singapore" "S2_Belinda" ...
##  $ Sample                 : int  6102461 6102465 6102466 6102467 6102470 6102475 6102478 6102479 6102481 6102486 ...
##  $ X                      : logi  NA NA NA NA NA NA ...
##  $ Category               : chr  "Healthy/Wild" "Healthy/Wild" "Healthy" "Healthy/Wild" ...
##  $ Sex.Age.Season         : chr  "FAP" "FAU" "FAA" "FAU" ...
##  $ Sex                    : chr  "F" "F" "F" "F" ...
##  $ Age                    : chr  "A" "A" "A" "A" ...
##  $ Season                 : Factor w/ 4 levels "Autumn","Spring",..: 2 3 1 3 3 4 1 1 1 4 ...
##  $ Location               : chr  "Channel" "Channel" "Channel" "Channel" ...
##  $ Agebyyear              : num  2.62 1.91 5 1.87 1.89 ...
##  $ Weight                 : chr  "7.2" "7.8" "8.6" "6.6" ...
##  $ X.1                    : logi  NA NA NA NA NA NA ...
##  $ Number.of.Tumours      : int  NA NA NA NA NA NA NA NA NA NA ...
##  $ Ulcerated              : logi  NA NA NA NA NA NA ...
##  $ UlcerationDepth        : chr  "" "" "" "" ...
##  $ Secondary.infection    : logi  NA NA NA NA NA NA ...
##  $ Total.tumour.volume    : num  NA NA NA NA NA NA NA NA NA NA ...
##  $ Barcode.sequence       : chr  "AGAGTAGC+AAGGCGTA" "GAAGGAAG+GACGTCAT" "ACAGCTCA+TACTGCTC" "GCAAGATC+AGTCGAAG" ...
##  $ PF.Clusters            : chr  "4,511,702" "6,479,047" "4,481,213" "3,296,547" ...
##  $ X..of.the.lane         : num  1 1 1 1 1 1 1 2 1 1 ...
##  $ X..Perfect.barcode     : num  98.3 97.3 98 97.4 95.7 ...
##  $ X..One.mismatch.barcode: num  1.71 2.72 2.01 2.6 4.3 3.33 2.2 2.25 2.23 2.19 ...
##  $ Yield..Mbases.         : chr  "1,363" "1,957" "1,353" "996" ...
##  $ X..PF.Cluster          : int  100 100 100 100 100 100 100 100 100 100 ...
##  $ X.....Q30              : num  93.8 93.8 94 94.1 93.9 ...
##  $ Mean.Quality           : num  35.9 35.9 36 36 35.9 ...
##  $ X.2                    : logi  NA NA NA NA NA NA ...
##  $ X.3                    : logi  NA NA NA NA NA NA ...
##  $ X.4                    : logi  NA NA NA NA NA NA ...
##  $ MvF                    : Factor w/ 2 levels "F","M": 1 1 1 1 1 1 2 2 2 2 ...
##  $ SubAdvAd               : num  2 2 2 2 2 2 2 2 2 2 ...

dds <- DESeqDataSetFromMatrix(countData = x1 , colData = ssHS, design = ~ Season +  SubAdvAd + MvF)

## converting counts to integer mode

##   the design formula contains one or more numeric variables with integer values,
##   specifying a model with increasing fold change for higher values.
##   did you mean for this to be a factor? if so, first convert
##   this variable to a factor using the factor() function

res <- DESeq(dds)

## estimating size factors

## estimating dispersions

## gene-wise dispersion estimates

## mean-dispersion relationship

## final dispersion estimates

## fitting model and testing

## -- replacing outliers and refitting for 10 genes
## -- DESeq argument 'minReplicatesForReplace' = 7 
## -- original counts are preserved in counts(dds)

## estimating dispersions

## fitting model and testing

z<- results(res)
vsd <- vst(dds, blind=FALSE,nsub=nrow(x1)-50)
zz<-cbind(as.data.frame(z),assay(vsd))
dge<-as.data.frame(zz[order(zz$pvalue),])
dgeHS <- dge
sig <- subset(dge,padj<0.05)
sig1_up <- rownames(subset(sig,log2FoldChange>0))
sig1_dn <- rownames(subset(sig,log2FoldChange<0))
length(sig1_up)

## [1] 0

maplot(dgeHS,"Male vs Female Healthy devils")

make_volcano(dgeHS,"Male vs Female Healthy devils")

col <- as.character(as.numeric(grepl("F",ssHS$Sex)))
col <- gsub("0","grey",col)
col <- gsub("1","black",col)
mtext("Sub-Adult=Orange, Adult=Brown")

sig[1:100,1:6] %>%
  kbl(caption="Comparison of Male vs Female Healthy devils") %>%
  kable_paper("hover", full_width = F)

Comparison of Male vs Female Healthy devils

	baseMean	log2FoldChange	lfcSE	stat	pvalue	padj
CXCLD2	23.75304	-4.970202	1.141820	-4.352876	1.34e-05	0.0034128
CCLD2	24.31172	-4.503590	1.108071	-4.064351	4.82e-05	0.0041248
CXCLD1	30.08543	-4.156034	1.023227	-4.061694	4.87e-05	0.0041248
NA	NA	NA	NA	NA	NA	NA
NA.1	NA	NA	NA	NA	NA	NA
NA.2	NA	NA	NA	NA	NA	NA
NA.3	NA	NA	NA	NA	NA	NA
NA.4	NA	NA	NA	NA	NA	NA
NA.5	NA	NA	NA	NA	NA	NA
NA.6	NA	NA	NA	NA	NA	NA
NA.7	NA	NA	NA	NA	NA	NA
NA.8	NA	NA	NA	NA	NA	NA
NA.9	NA	NA	NA	NA	NA	NA
NA.10	NA	NA	NA	NA	NA	NA
NA.11	NA	NA	NA	NA	NA	NA
NA.12	NA	NA	NA	NA	NA	NA
NA.13	NA	NA	NA	NA	NA	NA
NA.14	NA	NA	NA	NA	NA	NA
NA.15	NA	NA	NA	NA	NA	NA
NA.16	NA	NA	NA	NA	NA	NA
NA.17	NA	NA	NA	NA	NA	NA
NA.18	NA	NA	NA	NA	NA	NA
NA.19	NA	NA	NA	NA	NA	NA
NA.20	NA	NA	NA	NA	NA	NA
NA.21	NA	NA	NA	NA	NA	NA
NA.22	NA	NA	NA	NA	NA	NA
NA.23	NA	NA	NA	NA	NA	NA
NA.24	NA	NA	NA	NA	NA	NA
NA.25	NA	NA	NA	NA	NA	NA
NA.26	NA	NA	NA	NA	NA	NA
NA.27	NA	NA	NA	NA	NA	NA
NA.28	NA	NA	NA	NA	NA	NA
NA.29	NA	NA	NA	NA	NA	NA
NA.30	NA	NA	NA	NA	NA	NA
NA.31	NA	NA	NA	NA	NA	NA
NA.32	NA	NA	NA	NA	NA	NA
NA.33	NA	NA	NA	NA	NA	NA
NA.34	NA	NA	NA	NA	NA	NA
NA.35	NA	NA	NA	NA	NA	NA
NA.36	NA	NA	NA	NA	NA	NA
NA.37	NA	NA	NA	NA	NA	NA
NA.38	NA	NA	NA	NA	NA	NA
NA.39	NA	NA	NA	NA	NA	NA
NA.40	NA	NA	NA	NA	NA	NA
NA.41	NA	NA	NA	NA	NA	NA
NA.42	NA	NA	NA	NA	NA	NA
NA.43	NA	NA	NA	NA	NA	NA
NA.44	NA	NA	NA	NA	NA	NA
NA.45	NA	NA	NA	NA	NA	NA
NA.46	NA	NA	NA	NA	NA	NA
NA.47	NA	NA	NA	NA	NA	NA
NA.48	NA	NA	NA	NA	NA	NA
NA.49	NA	NA	NA	NA	NA	NA
NA.50	NA	NA	NA	NA	NA	NA
NA.51	NA	NA	NA	NA	NA	NA
NA.52	NA	NA	NA	NA	NA	NA
NA.53	NA	NA	NA	NA	NA	NA
NA.54	NA	NA	NA	NA	NA	NA
NA.55	NA	NA	NA	NA	NA	NA
NA.56	NA	NA	NA	NA	NA	NA
NA.57	NA	NA	NA	NA	NA	NA
NA.58	NA	NA	NA	NA	NA	NA
NA.59	NA	NA	NA	NA	NA	NA
NA.60	NA	NA	NA	NA	NA	NA
NA.61	NA	NA	NA	NA	NA	NA
NA.62	NA	NA	NA	NA	NA	NA
NA.63	NA	NA	NA	NA	NA	NA
NA.64	NA	NA	NA	NA	NA	NA
NA.65	NA	NA	NA	NA	NA	NA
NA.66	NA	NA	NA	NA	NA	NA
NA.67	NA	NA	NA	NA	NA	NA
NA.68	NA	NA	NA	NA	NA	NA
NA.69	NA	NA	NA	NA	NA	NA
NA.70	NA	NA	NA	NA	NA	NA
NA.71	NA	NA	NA	NA	NA	NA
NA.72	NA	NA	NA	NA	NA	NA
NA.73	NA	NA	NA	NA	NA	NA
NA.74	NA	NA	NA	NA	NA	NA
NA.75	NA	NA	NA	NA	NA	NA
NA.76	NA	NA	NA	NA	NA	NA
NA.77	NA	NA	NA	NA	NA	NA
NA.78	NA	NA	NA	NA	NA	NA
NA.79	NA	NA	NA	NA	NA	NA
NA.80	NA	NA	NA	NA	NA	NA
NA.81	NA	NA	NA	NA	NA	NA
NA.82	NA	NA	NA	NA	NA	NA
NA.83	NA	NA	NA	NA	NA	NA
NA.84	NA	NA	NA	NA	NA	NA
NA.85	NA	NA	NA	NA	NA	NA
NA.86	NA	NA	NA	NA	NA	NA
NA.87	NA	NA	NA	NA	NA	NA
NA.88	NA	NA	NA	NA	NA	NA
NA.89	NA	NA	NA	NA	NA	NA
NA.90	NA	NA	NA	NA	NA	NA
NA.91	NA	NA	NA	NA	NA	NA
NA.92	NA	NA	NA	NA	NA	NA
NA.93	NA	NA	NA	NA	NA	NA
NA.94	NA	NA	NA	NA	NA	NA
NA.95	NA	NA	NA	NA	NA	NA
NA.96	NA	NA	NA	NA	NA	NA

write.table(dge,file="dgeHS.tsv",sep="\t")

mx <- sig[,7:ncol(sig)]
mx <- head(mx,30)
colfunc <- colorRampPalette(c("blue", "white", "red"))
heatmap.2(as.matrix(mx),trace="none",scale="row",
          col=colfunc(25),ColSideColors=col,mar=c(5,12))

ssDFT <- filter(ss, Category %in% c('DFT1','DFT2'))

ssDFTS <- ssDFT[which(grepl("M",ssDFT$Sex) + grepl("F",ssDFT$Sex)>0),]
ssDFTS

##    Lane Project              Client.ID  Sample  X Category Sex.Age.Season Sex
## 1     1 default            S1_Sarafina 6102463 NA     DFT1            FAP   F
## 2     1 default                S1_Lucy 6102464 NA     DFT1            FAU   F
## 3     1 default                S2_Hope 6102471 NA     DFT1            FAU   F
## 4     1 default               S2_Sally 6102472 NA     DFT1            FAU   F
## 5     1 default                 S3_Apu 6102473 NA     DFT1            MAA   M
## 6     1 default              S3_Simone 6102474 NA     DFT1            FAW   F
## 7     1 default              S4_Wiggum 6102480 NA     DFT1            MAA   M
## 8     1 default                S4_Pooh 6102482 NA     DFT1            MAA   M
## 9     1 default            S4_Milhouse 6102483 NA     DFT1            MAA   M
## 10    1 default               S5_Allen 6102488 NA     DFT1            MAU   M
## 11    1 default                S5_Cyan 6102489 NA     DFT1            MAW   M
## 12    1 default              S6_Shelob 6102492 NA     DFT2            FAP   F
## 13    1 default           S6_Stuttgart 6102493 NA     DFT1            MSP   M
## 14    1 default              S7_Violet 6102498 NA     DFT2            FSU   F
## 15    1 default              S7_Willie 6102500 NA     DFT2            MAA   M
## 16    1 default              S8_Orange 6102502 NA     DFT1            FAU   F
## 17    1 default                S8_ST67 6102505 NA     DFT2            MAW   M
## 18    1 default              S8_Yellow 6102507 NA     DFT2            MAU   M
## 19    1 default                S9_Jane 6102508 NA     DFT1            FAA   F
## 20    1 default        S9_Clytemnestra 6102511 NA     DFT1            FSP   F
## 21    1 default                S9_Zeus 6102512 NA     DFT2            MSW   M
## 22    1 default           S10_Laurasia 6102516 NA     DFT2            MSP   M
## 23    1 default          S10_Enchilada 6102517 NA     DFT2            MAP   M
## 24    1 default S11_Murdoch/Washington 6102519 NA     DFT2            MAA   M
## 25    1 default                S12_Abe 6102524 NA     DFT1            MAA   M
## 26    1 default              S12_Lilac 6102526 NA     DFT1            MAW   M
## 27    1 default              S13_Green 6102531 NA     DFT1            MAA   M
## 28    1 default               S13_Bump 6102532 NA     DFT1            MAU   M
## 29    1 default             S14_Purple 6102536 NA     DFT1            FAW   F
##    Age Season Location Agebyyear Weight X.1 Number.of.Tumours Ulcerated
## 1    A Spring  Channel 2.6191781    6.4  NA                NA        NA
## 2    A Summer  Channel 2.9178082    7.1  NA                 3      TRUE
## 3    A Summer  Channel 2.9315068    6.4  NA                 4      TRUE
## 4    A Summer  Channel 2.9945205      7  NA                 1      TRUE
## 5    A Autumn  Channel 3.1150685    9.5  NA                 1     FALSE
## 6    A Winter  Channel 3.2739726   10.2  NA                 1      TRUE
## 7    A Autumn  Channel 3.1178082    8.3  NA                 2     FALSE
## 8    A Autumn  Channel 4.1863014   11.8  NA                 1     FALSE
## 9    A Autumn  Channel 4.1013699     11  NA                 1      TRUE
## 10   A Summer  Channel 1.9095890    6.7  NA                 1      TRUE
## 11   A Winter  Channel 3.3945205    9.1  NA                 5      TRUE
## 12   A Spring  Channel 4.6219178    7.2  NA                NA        NA
## 13   S Spring  Channel 0.5780822   12.7  NA                 1     FALSE
## 14   S Summer  Channel 0.8739726    4.6  NA                 1      TRUE
## 15   A Autumn  Channel 2.1150685    8.2  NA                 2      TRUE
## 16   A Summer  Channel 1.9150685    7.1  NA                 7      TRUE
## 17   A Winter  Channel 4.2794521   10.2  NA                 1      TRUE
## 18   A Summer  Channel 1.8657534   10.5  NA                 2      TRUE
## 19   A Autumn  Channel 3.1972603    6.1  NA                NA        NA
## 20   S Spring  Channel 1.6602740      6  NA                NA        NA
## 21   S Winter  Channel 1.3479452    7.3  NA                NA        NA
## 22   S Spring  Channel 1.6383562      8  NA                NA        NA
## 23   A Spring  Channel 3.6438356    9.3  NA                14     FALSE
## 24   A Autumn  Channel 1.4438356    8.9  NA                NA        NA
## 25   A Autumn  Channel 3.1150685    7.9  NA                 1      TRUE
## 26   A Winter  Channel 3.4109589    9.6  NA                 5      TRUE
## 27   A Autumn  Channel 3.1534247    7.5  NA                 2      TRUE
## 28   A Summer  Channel 3.9068493    9.3  NA                NA        NA
## 29   A Winter  Channel 2.3945205    6.8  NA                 1      TRUE
##    UlcerationDepth Secondary.infection Total.tumour.volume  Barcode.sequence
## 1                                   NA           95140.790 ATCCAGAG+TAACGTCG
## 2             deep                TRUE          700825.170 AAGCACTG+AGGTCAAC
## 3             deep                TRUE          103815.530 AGCACTTC+ACGTCCAA
## 4             deep               FALSE            1588.272 CGTTGCAA+ATAGAGCG
## 5      superficial                TRUE             200.000 CCTTGATC+GATGGAGT
## 6      superficial                TRUE            2670.678 CGGTTGTT+GTGGTATG
## 7                                FALSE            7272.309 GTGAAGTG+GAGCAATC
## 8                                FALSE            7577.856 CTTCGACT+GATCTTGC
## 9             deep                TRUE             480.000 AACCGTTC+CTAGCTCA
## 10            deep                TRUE           55959.750 TTGCAGAC+ATCGTGGT
## 11            deep                TRUE          206006.422 GTAGGAGT+GTCCTAAG
## 12                                  NA            7455.435 GTCATCGA+GATCCACT
## 13                                TRUE            3809.806 AGCGTGTT+CTTCGCAA
## 14     superficial                TRUE            5517.330 CTTGTCGA+CGATGTTC
## 15            deep                TRUE            1527.480 TGGACTCT+TCTTACGG
## 16     superficial                TRUE           43690.461 GTTGACCT+CAGTGCTT
## 17            deep                TRUE          102246.432 TTGGACGT+GAAGTGCT
## 18            deep                TRUE          139389.671 GAAGTTGG+CGAACAAC
## 19                                  NA           64646.036 GTTCTCGT+TGTTCCGT
## 20                                  NA           34738.080 CATGGCTA+CACACATC
## 21                                  NA             100.000 CGTGTGTA+CCAACGAA
## 22                                  NA            2697.152 TACGCTAC+CGTGTGAT
## 23                               FALSE          301301.905 GATTGCTC+CACTTCAC
## 24                                  NA          127176.305 ACTGAGGT+GTCAACAG
## 25            deep                TRUE           13125.000 GTTGTTCG+CCAACTTC
## 26     superficial                TRUE           37293.844 TATCGGTC+TACCGGAT
## 27     superficial                TRUE            7407.222 GTCGAAGA+CCACATTG
## 28                                  NA          629522.362 AGTTCGTC+ACTCGATC
## 29            deep                TRUE           24938.123 CGCTCTAT+TTGCAACG
##    PF.Clusters X..of.the.lane X..Perfect.barcode X..One.mismatch.barcode
## 1    6,491,488              1              74.07                   25.93
## 2    6,121,785              1              98.00                    2.00
## 3    1,971,776              0              97.96                    2.04
## 4    2,429,449              1              97.42                    2.58
## 5    4,939,580              1              96.94                    3.06
## 6    4,015,027              1              95.91                    4.09
## 7    4,201,355              1              97.17                    2.83
## 8    3,821,151              1              97.31                    2.69
## 9   10,492,900              2              98.13                    1.87
## 10   5,288,435              1              96.85                    3.15
## 11   4,138,600              1              98.03                    1.97
## 12   3,184,798              1              98.21                    1.79
## 13   3,056,475              1              97.50                    2.50
## 14   2,768,682              1              98.19                    1.81
## 15   4,539,036              1              98.30                    1.70
## 16   5,072,309              1              97.92                    2.08
## 17   5,476,735              1              95.44                    4.56
## 18   6,469,324              1              97.99                    2.01
## 19  12,489,105              3              97.92                    2.08
## 20   6,141,790              1              98.00                    2.00
## 21     699,953              0              97.64                    2.36
## 22   9,716,148              2              97.81                    2.19
## 23  11,967,844              2              97.82                    2.18
## 24   2,003,509              0              97.82                    2.18
## 25   6,446,467              1              97.60                    2.40
## 26   4,490,241              1              98.15                    1.85
## 27   2,623,695              1              97.38                    2.62
## 28   9,435,053              2              98.16                    1.84
## 29   5,432,182              1              97.81                    2.19
##    Yield..Mbases. X..PF.Cluster X.....Q30 Mean.Quality X.2 X.3 X.4
## 1           1,960           100     93.74        35.92  NA  NA  NA
## 2           1,849           100     93.84        35.93  NA  NA  NA
## 3             595           100     93.62        35.88  NA  NA  NA
## 4             734           100     93.78        35.93  NA  NA  NA
## 5           1,492           100     93.71        35.91  NA  NA  NA
## 6           1,213           100     93.42        35.86  NA  NA  NA
## 7           1,269           100     93.64        35.90  NA  NA  NA
## 8           1,154           100     93.60        35.89  NA  NA  NA
## 9           3,169           100     93.80        35.92  NA  NA  NA
## 10          1,597           100     93.42        35.86  NA  NA  NA
## 11          1,250           100     93.67        35.90  NA  NA  NA
## 12            962           100     93.90        35.94  NA  NA  NA
## 13            923           100     93.95        35.95  NA  NA  NA
## 14            836           100     93.85        35.93  NA  NA  NA
## 15          1,371           100     93.84        35.93  NA  NA  NA
## 16          1,532           100     93.36        35.84  NA  NA  NA
## 17          1,654           100     93.21        35.82  NA  NA  NA
## 18          1,954           100     94.00        35.97  NA  NA  NA
## 19          3,772           100     93.54        35.88  NA  NA  NA
## 20          1,855           100     93.42        35.85  NA  NA  NA
## 21            211           100     93.33        35.84  NA  NA  NA
## 22          2,934           100     93.97        35.95  NA  NA  NA
## 23          3,614           100     93.88        35.94  NA  NA  NA
## 24            605           100     93.39        35.85  NA  NA  NA
## 25          1,947           100     93.65        35.90  NA  NA  NA
## 26          1,356           100     93.42        35.85  NA  NA  NA
## 27            792           100     92.94        35.76  NA  NA  NA
## 28          2,849           100     93.64        35.89  NA  NA  NA
## 29          1,641           100     93.93        35.95  NA  NA  NA

ssDFTS <- ssDFTS[ssDFTS$Sample %in% colnames(x),]

dim(ssDFTS)

## [1] 28 31

x1 <- x[,which(colnames(x) %in% ssDFTS$Sample)]
dim(x1)

## [1] 310  28

x1 <- x1[which(rowMeans(x1)>10),]
dim(x1)

## [1] 258  28

ssDFTS %>%  kbl(caption="Sample sheet")%>% kable_styling("hover",full_width=FALSE) 

Sample sheet

	Lane	Project	Client.ID	Sample	X	Category	Sex.Age.Season	Sex	Age	Season	Location	Agebyyear	Weight	X.1	Number.of.Tumours	Ulcerated	UlcerationDepth	Secondary.infection	Total.tumour.volume	Barcode.sequence	PF.Clusters	X..of.the.lane	X..Perfect.barcode	X..One.mismatch.barcode	Yield..Mbases.	X..PF.Cluster	X…..Q30	Mean.Quality	X.2	X.3	X.4
1	1	default	S1_Sarafina	6102463	NA	DFT1	FAP	F	A	Spring	Channel	2.6191781	6.4	NA	NA	NA		NA	95140.790	ATCCAGAG+TAACGTCG	6,491,488	1	74.07	25.93	1,960	100	93.74	35.92	NA	NA	NA
2	1	default	S1_Lucy	6102464	NA	DFT1	FAU	F	A	Summer	Channel	2.9178082	7.1	NA	3	TRUE	deep	TRUE	700825.170	AAGCACTG+AGGTCAAC	6,121,785	1	98.00	2.00	1,849	100	93.84	35.93	NA	NA	NA
3	1	default	S2_Hope	6102471	NA	DFT1	FAU	F	A	Summer	Channel	2.9315068	6.4	NA	4	TRUE	deep	TRUE	103815.530	AGCACTTC+ACGTCCAA	1,971,776	0	97.96	2.04	595	100	93.62	35.88	NA	NA	NA
4	1	default	S2_Sally	6102472	NA	DFT1	FAU	F	A	Summer	Channel	2.9945205	7	NA	1	TRUE	deep	FALSE	1588.272	CGTTGCAA+ATAGAGCG	2,429,449	1	97.42	2.58	734	100	93.78	35.93	NA	NA	NA
5	1	default	S3_Apu	6102473	NA	DFT1	MAA	M	A	Autumn	Channel	3.1150685	9.5	NA	1	FALSE	superficial	TRUE	200.000	CCTTGATC+GATGGAGT	4,939,580	1	96.94	3.06	1,492	100	93.71	35.91	NA	NA	NA
6	1	default	S3_Simone	6102474	NA	DFT1	FAW	F	A	Winter	Channel	3.2739726	10.2	NA	1	TRUE	superficial	TRUE	2670.678	CGGTTGTT+GTGGTATG	4,015,027	1	95.91	4.09	1,213	100	93.42	35.86	NA	NA	NA
7	1	default	S4_Wiggum	6102480	NA	DFT1	MAA	M	A	Autumn	Channel	3.1178082	8.3	NA	2	FALSE		FALSE	7272.309	GTGAAGTG+GAGCAATC	4,201,355	1	97.17	2.83	1,269	100	93.64	35.90	NA	NA	NA
8	1	default	S4_Pooh	6102482	NA	DFT1	MAA	M	A	Autumn	Channel	4.1863014	11.8	NA	1	FALSE		FALSE	7577.856	CTTCGACT+GATCTTGC	3,821,151	1	97.31	2.69	1,154	100	93.60	35.89	NA	NA	NA
9	1	default	S4_Milhouse	6102483	NA	DFT1	MAA	M	A	Autumn	Channel	4.1013699	11	NA	1	TRUE	deep	TRUE	480.000	AACCGTTC+CTAGCTCA	10,492,900	2	98.13	1.87	3,169	100	93.80	35.92	NA	NA	NA
10	1	default	S5_Allen	6102488	NA	DFT1	MAU	M	A	Summer	Channel	1.9095890	6.7	NA	1	TRUE	deep	TRUE	55959.750	TTGCAGAC+ATCGTGGT	5,288,435	1	96.85	3.15	1,597	100	93.42	35.86	NA	NA	NA
11	1	default	S5_Cyan	6102489	NA	DFT1	MAW	M	A	Winter	Channel	3.3945205	9.1	NA	5	TRUE	deep	TRUE	206006.422	GTAGGAGT+GTCCTAAG	4,138,600	1	98.03	1.97	1,250	100	93.67	35.90	NA	NA	NA
12	1	default	S6_Shelob	6102492	NA	DFT2	FAP	F	A	Spring	Channel	4.6219178	7.2	NA	NA	NA		NA	7455.435	GTCATCGA+GATCCACT	3,184,798	1	98.21	1.79	962	100	93.90	35.94	NA	NA	NA
13	1	default	S6_Stuttgart	6102493	NA	DFT1	MSP	M	S	Spring	Channel	0.5780822	12.7	NA	1	FALSE		TRUE	3809.806	AGCGTGTT+CTTCGCAA	3,056,475	1	97.50	2.50	923	100	93.95	35.95	NA	NA	NA
14	1	default	S7_Violet	6102498	NA	DFT2	FSU	F	S	Summer	Channel	0.8739726	4.6	NA	1	TRUE	superficial	TRUE	5517.330	CTTGTCGA+CGATGTTC	2,768,682	1	98.19	1.81	836	100	93.85	35.93	NA	NA	NA
15	1	default	S7_Willie	6102500	NA	DFT2	MAA	M	A	Autumn	Channel	2.1150685	8.2	NA	2	TRUE	deep	TRUE	1527.480	TGGACTCT+TCTTACGG	4,539,036	1	98.30	1.70	1,371	100	93.84	35.93	NA	NA	NA
16	1	default	S8_Orange	6102502	NA	DFT1	FAU	F	A	Summer	Channel	1.9150685	7.1	NA	7	TRUE	superficial	TRUE	43690.461	GTTGACCT+CAGTGCTT	5,072,309	1	97.92	2.08	1,532	100	93.36	35.84	NA	NA	NA
17	1	default	S8_ST67	6102505	NA	DFT2	MAW	M	A	Winter	Channel	4.2794521	10.2	NA	1	TRUE	deep	TRUE	102246.432	TTGGACGT+GAAGTGCT	5,476,735	1	95.44	4.56	1,654	100	93.21	35.82	NA	NA	NA
18	1	default	S8_Yellow	6102507	NA	DFT2	MAU	M	A	Summer	Channel	1.8657534	10.5	NA	2	TRUE	deep	TRUE	139389.671	GAAGTTGG+CGAACAAC	6,469,324	1	97.99	2.01	1,954	100	94.00	35.97	NA	NA	NA
19	1	default	S9_Jane	6102508	NA	DFT1	FAA	F	A	Autumn	Channel	3.1972603	6.1	NA	NA	NA		NA	64646.036	GTTCTCGT+TGTTCCGT	12,489,105	3	97.92	2.08	3,772	100	93.54	35.88	NA	NA	NA
20	1	default	S9_Clytemnestra	6102511	NA	DFT1	FSP	F	S	Spring	Channel	1.6602740	6	NA	NA	NA		NA	34738.080	CATGGCTA+CACACATC	6,141,790	1	98.00	2.00	1,855	100	93.42	35.85	NA	NA	NA
22	1	default	S10_Laurasia	6102516	NA	DFT2	MSP	M	S	Spring	Channel	1.6383562	8	NA	NA	NA		NA	2697.152	TACGCTAC+CGTGTGAT	9,716,148	2	97.81	2.19	2,934	100	93.97	35.95	NA	NA	NA
23	1	default	S10_Enchilada	6102517	NA	DFT2	MAP	M	A	Spring	Channel	3.6438356	9.3	NA	14	FALSE		FALSE	301301.905	GATTGCTC+CACTTCAC	11,967,844	2	97.82	2.18	3,614	100	93.88	35.94	NA	NA	NA
24	1	default	S11_Murdoch/Washington	6102519	NA	DFT2	MAA	M	A	Autumn	Channel	1.4438356	8.9	NA	NA	NA		NA	127176.305	ACTGAGGT+GTCAACAG	2,003,509	0	97.82	2.18	605	100	93.39	35.85	NA	NA	NA
25	1	default	S12_Abe	6102524	NA	DFT1	MAA	M	A	Autumn	Channel	3.1150685	7.9	NA	1	TRUE	deep	TRUE	13125.000	GTTGTTCG+CCAACTTC	6,446,467	1	97.60	2.40	1,947	100	93.65	35.90	NA	NA	NA
26	1	default	S12_Lilac	6102526	NA	DFT1	MAW	M	A	Winter	Channel	3.4109589	9.6	NA	5	TRUE	superficial	TRUE	37293.844	TATCGGTC+TACCGGAT	4,490,241	1	98.15	1.85	1,356	100	93.42	35.85	NA	NA	NA
27	1	default	S13_Green	6102531	NA	DFT1	MAA	M	A	Autumn	Channel	3.1534247	7.5	NA	2	TRUE	superficial	TRUE	7407.222	GTCGAAGA+CCACATTG	2,623,695	1	97.38	2.62	792	100	92.94	35.76	NA	NA	NA
28	1	default	S13_Bump	6102532	NA	DFT1	MAU	M	A	Summer	Channel	3.9068493	9.3	NA	NA	NA		NA	629522.362	AGTTCGTC+ACTCGATC	9,435,053	2	98.16	1.84	2,849	100	93.64	35.89	NA	NA	NA
29	1	default	S14_Purple	6102536	NA	DFT1	FAW	F	A	Winter	Channel	2.3945205	6.8	NA	1	TRUE	deep	TRUE	24938.123	CGCTCTAT+TTGCAACG	5,432,182	1	97.81	2.19	1,641	100	93.93	35.95	NA	NA	NA

col <- as.character(as.numeric(grepl("F",ssHS$Sex)))
col <- gsub("0","grey",col)
col <- gsub("1","black",col)

mds<-plotMDS(x1,pch=19,cex=3,col=col,main="MDS plot: Comparison of Male and Female DFTD infected devils")
mtext("Male=Grey, Female=Black")

ssDFTS$DFTDStat <- grepl("DFT2",ssDFTS$Category)
ssDFTS$MvF <- factor(ssDFTS$Sex)
ssDFTS$SubAdvAd <- as.numeric(factor(ssDFTS$Age,levels=c("S","A"),ordered = TRUE))
ssDFTS$Season <- factor(ssDFTS$Season)


dds <- DESeqDataSetFromMatrix(countData = x1 , colData = ssDFTS, design = ~ Season +  SubAdvAd + DFTDStat + MvF)

## converting counts to integer mode

##   the design formula contains one or more numeric variables with integer values,
##   specifying a model with increasing fold change for higher values.
##   did you mean for this to be a factor? if so, first convert
##   this variable to a factor using the factor() function

res <- DESeq(dds)

## estimating size factors

## estimating dispersions

## gene-wise dispersion estimates

## mean-dispersion relationship

## final dispersion estimates

## fitting model and testing

z<- results(res)
vsd <- vst(dds, blind=FALSE,nsub=nrow(x1)-50)
zz<-cbind(as.data.frame(z),assay(vsd))
dge<-as.data.frame(zz[order(zz$pvalue),])
dgeDFTS <- dge
sig <- subset(dge,padj<0.05)
sig1_up <- rownames(subset(sig,log2FoldChange>0))
sig1_dn <- rownames(subset(sig,log2FoldChange<0))
length(sig1_up)

## [1] 1

maplot(dgeDFTS,"Male vs Female DFTD infected devils")

make_volcano(dgeDFTS,"Male vs Female DFTD infected devils")

sig[1:100,1:6] %>%
  kbl(caption="Comparison of Male vs Female DFTD infected devils") %>%
  kable_paper("hover", full_width = F)

Comparison of Male vs Female DFTD infected devils

	baseMean	log2FoldChange	lfcSE	stat	pvalue	padj
IL9R	22.190197	-4.778219	1.159975	-4.119242	0.0000380	0.0098071
CXCL10	6.852306	4.868341	1.356887	3.587875	0.0003334	0.0430066
NA	NA	NA	NA	NA	NA	NA
NA.1	NA	NA	NA	NA	NA	NA
NA.2	NA	NA	NA	NA	NA	NA
NA.3	NA	NA	NA	NA	NA	NA
NA.4	NA	NA	NA	NA	NA	NA
NA.5	NA	NA	NA	NA	NA	NA
NA.6	NA	NA	NA	NA	NA	NA
NA.7	NA	NA	NA	NA	NA	NA
NA.8	NA	NA	NA	NA	NA	NA
NA.9	NA	NA	NA	NA	NA	NA
NA.10	NA	NA	NA	NA	NA	NA
NA.11	NA	NA	NA	NA	NA	NA
NA.12	NA	NA	NA	NA	NA	NA
NA.13	NA	NA	NA	NA	NA	NA
NA.14	NA	NA	NA	NA	NA	NA
NA.15	NA	NA	NA	NA	NA	NA
NA.16	NA	NA	NA	NA	NA	NA
NA.17	NA	NA	NA	NA	NA	NA
NA.18	NA	NA	NA	NA	NA	NA
NA.19	NA	NA	NA	NA	NA	NA
NA.20	NA	NA	NA	NA	NA	NA
NA.21	NA	NA	NA	NA	NA	NA
NA.22	NA	NA	NA	NA	NA	NA
NA.23	NA	NA	NA	NA	NA	NA
NA.24	NA	NA	NA	NA	NA	NA
NA.25	NA	NA	NA	NA	NA	NA
NA.26	NA	NA	NA	NA	NA	NA
NA.27	NA	NA	NA	NA	NA	NA
NA.28	NA	NA	NA	NA	NA	NA
NA.29	NA	NA	NA	NA	NA	NA
NA.30	NA	NA	NA	NA	NA	NA
NA.31	NA	NA	NA	NA	NA	NA
NA.32	NA	NA	NA	NA	NA	NA
NA.33	NA	NA	NA	NA	NA	NA
NA.34	NA	NA	NA	NA	NA	NA
NA.35	NA	NA	NA	NA	NA	NA
NA.36	NA	NA	NA	NA	NA	NA
NA.37	NA	NA	NA	NA	NA	NA
NA.38	NA	NA	NA	NA	NA	NA
NA.39	NA	NA	NA	NA	NA	NA
NA.40	NA	NA	NA	NA	NA	NA
NA.41	NA	NA	NA	NA	NA	NA
NA.42	NA	NA	NA	NA	NA	NA
NA.43	NA	NA	NA	NA	NA	NA
NA.44	NA	NA	NA	NA	NA	NA
NA.45	NA	NA	NA	NA	NA	NA
NA.46	NA	NA	NA	NA	NA	NA
NA.47	NA	NA	NA	NA	NA	NA
NA.48	NA	NA	NA	NA	NA	NA
NA.49	NA	NA	NA	NA	NA	NA
NA.50	NA	NA	NA	NA	NA	NA
NA.51	NA	NA	NA	NA	NA	NA
NA.52	NA	NA	NA	NA	NA	NA
NA.53	NA	NA	NA	NA	NA	NA
NA.54	NA	NA	NA	NA	NA	NA
NA.55	NA	NA	NA	NA	NA	NA
NA.56	NA	NA	NA	NA	NA	NA
NA.57	NA	NA	NA	NA	NA	NA
NA.58	NA	NA	NA	NA	NA	NA
NA.59	NA	NA	NA	NA	NA	NA
NA.60	NA	NA	NA	NA	NA	NA
NA.61	NA	NA	NA	NA	NA	NA
NA.62	NA	NA	NA	NA	NA	NA
NA.63	NA	NA	NA	NA	NA	NA
NA.64	NA	NA	NA	NA	NA	NA
NA.65	NA	NA	NA	NA	NA	NA
NA.66	NA	NA	NA	NA	NA	NA
NA.67	NA	NA	NA	NA	NA	NA
NA.68	NA	NA	NA	NA	NA	NA
NA.69	NA	NA	NA	NA	NA	NA
NA.70	NA	NA	NA	NA	NA	NA
NA.71	NA	NA	NA	NA	NA	NA
NA.72	NA	NA	NA	NA	NA	NA
NA.73	NA	NA	NA	NA	NA	NA
NA.74	NA	NA	NA	NA	NA	NA
NA.75	NA	NA	NA	NA	NA	NA
NA.76	NA	NA	NA	NA	NA	NA
NA.77	NA	NA	NA	NA	NA	NA
NA.78	NA	NA	NA	NA	NA	NA
NA.79	NA	NA	NA	NA	NA	NA
NA.80	NA	NA	NA	NA	NA	NA
NA.81	NA	NA	NA	NA	NA	NA
NA.82	NA	NA	NA	NA	NA	NA
NA.83	NA	NA	NA	NA	NA	NA
NA.84	NA	NA	NA	NA	NA	NA
NA.85	NA	NA	NA	NA	NA	NA
NA.86	NA	NA	NA	NA	NA	NA
NA.87	NA	NA	NA	NA	NA	NA
NA.88	NA	NA	NA	NA	NA	NA
NA.89	NA	NA	NA	NA	NA	NA
NA.90	NA	NA	NA	NA	NA	NA
NA.91	NA	NA	NA	NA	NA	NA
NA.92	NA	NA	NA	NA	NA	NA
NA.93	NA	NA	NA	NA	NA	NA
NA.94	NA	NA	NA	NA	NA	NA
NA.95	NA	NA	NA	NA	NA	NA
NA.96	NA	NA	NA	NA	NA	NA
NA.97	NA	NA	NA	NA	NA	NA

write.table(dge,file="dgeDFTS.tsv",sep="\t")

ssCap <- filter(ss, Category %in% c('Captive'))

ssCapS <- ssCap[which(grepl("M",ssCap$Sex) + grepl("F",ssCap$Sex)>0),]
ssCapS

##    Lane Project      Client.ID  Sample  X Category Sex.Age.Season Sex Age
## 1     1 default      S2_B99008 6102468 NA  Captive            FAU   F   A
## 2     1 default      S2_Bonney 6102469 NA  Captive            FAU   F   A
## 3     1 default      S3_Jessie 6102476 NA  Captive            FAW   F   A
## 4     1 default    S3_McCauley 6102477 NA  Captive            MAA   M   A
## 5     1 default      S5_Fenton 6102484 NA  Captive            MAU   M   A
## 6     1 default S5_Bryan Adams 6102485 NA  Captive            MAW   M   A
## 7     1 default       S7_Fredo 6102496 NA  Captive            MAA   M   A
## 8     1 default      S7_Piglet 6102497 NA  Captive            FSU   F   S
## 9     1 default    S10_McMahon 6102514 NA  Captive            MSP   M   S
## 10    1 default      S10_Logan 6102515 NA  Captive            MAU   M   A
## 11    1 default       S11_Xuki 6102520 NA  Captive            MAA   M   A
## 12    1 default   S11_Kalamata 6102521 NA  Captive            MSW   M   S
## 13    1 default      S12_Wally 6102528 NA  Captive            MAA   M   A
## 14    1 default        S12_Naz 6102529 NA  Captive            MAW   M   A
## 15    1 default     S13_Humbug 6102530 NA  Captive            MAU   M   A
## 16    1 default     S13_Xerxes 6102535 NA  Captive            MAA   M   A
## 17    1 default     S14_Catani 6102538 NA  Captive            FAU   F   A
## 18    2 default     S15_Amelia 6102541 NA  Captive            FAU   F   A
## 19    2 default     S15_London 6102542 NA  Captive            FAU   F   A
## 20    2 default      S15_Elyse 6102543 NA  Captive            FAP   F   A
## 21    2 default     S16_Marley 6102547 NA  Captive            FAU   F   A
## 22    2 default      S16_Milla 6102548 NA  Captive            FAU   F   A
## 23    2 default S16_Maleficent 6102549 NA  Captive            FAU   F   A
## 24    2 default     S17_Anirak 6102553 NA  Captive            FSW   F   S
## 25    2 default       S17_Poly 6102554 NA  Captive            FSU   F   S
## 26    2 default     S17_Methyl 6102556 NA  Captive            FSU   F   S
## 27    2 default       S18_Zara 6102558 NA  Captive            FSW   F   S
## 28    2 default       S18_Maze 6102559 NA  Captive            FSW   F   S
## 29    2 default    S18_Delores 6102560 NA  Captive            FSW   F   S
## 30    2 default    S19_Morocco 6102564 NA  Captive            MAU   M   A
## 31    2 default   S19_Pecorino 6102565 NA  Captive            FSW   F   S
## 32    2 default      S19_Hooch 6102566 NA  Captive            MSA   M   S
## 33    2 default      S20_Muasy 6102569 NA  Captive            MSU   M   S
## 34    2 default       S20_Fuse 6102570 NA  Captive            MSU   M   S
## 35    2 default     S20_Joseph 6102571 NA  Captive            MSU   M   S
## 36    2 default      S21_Moama 6102576 NA  Captive            MAW   M   A
## 37    2 default     S21_Nauset 6102577 NA  Captive            MSW   M   S
## 38    2 default     S21_Graham 6102578 NA  Captive            MSW   M   S
## 39    2 default      S22_Yanni 6102581 NA  Captive            MSW   M   S
## 40    2 default    S22_Gherkin 6102582 NA  Captive            MSW   M   S
## 41    2 default    S22_Stilton 6102584 NA  Captive            MSW   M   S
## 42    2 default     S23_Osiris 6102588 NA  Captive            MSW   M   S
## 43    2 default     S23_X-boxx 6102589 NA  Captive            FSW   F   S
## 44    2 default     S23_Quince 6102590 NA  Captive            MSW   M   S
## 45    2 default     S24_Malfoy 6102595 NA  Captive            MAW   M   A
## 46    2 default   S24_Licorice 6102596 NA  Captive            FSA   F   S
## 47    2 default       S25_Tofu 6102599 NA  Captive            FAU   F   A
## 48    2 default    S25_Colette 6102600 NA  Captive            FAP   F   A
## 49    2 default S25_CannonBall 6102601 NA  Captive            MSW   M   S
## 50    2 default  S25_Cornelius 6102602 NA  Captive            MAU   M   A
## 51    2 default    S25_Mardook 6102603 NA  Captive            MAW   M   A
## 52    2 default    S25_Negroni 6102604 NA  Captive            FSP   F   S
##    Season    Location Agebyyear Weight X.1 Number.of.Tumours Ulcerated
## 1  Summer     Monarto 3.7369863   8.12  NA                NA        NA
## 2  Summer Healesville 2.5863014    5.4  NA                NA        NA
## 3  Winter     Monarto 2.1945205  12.95  NA                NA        NA
## 4  Autumn Healesville 3.0328767   8.52  NA                NA        NA
## 5  Summer     Monarto 3.0000000    N/A  NA                NA        NA
## 6  Winter Healesville 4.1616438    8.6  NA                NA        NA
## 7  Autumn Healesville 2.9945205    8.8  NA                NA        NA
## 8  Summer Healesville 1.5890411   4.76  NA                NA        NA
## 9  Spring Healesville 1.6493151   8.06  NA                NA        NA
## 10 Summer Healesville 5.6000000   9.02  NA                NA        NA
## 11 Autumn     Monarto 2.0767123   7.95  NA                NA        NA
## 12 Winter Healesville 1.2849315   5.58  NA                NA        NA
## 13 Autumn     Monarto 2.1616438    N/A  NA                NA        NA
## 14 Winter     Monarto 4.4547945    N/A  NA                NA        NA
## 15 Summer     Monarto 2.5397260   7.94  NA                NA        NA
## 16 Autumn     Monarto 2.0767123    N/A  NA                NA        NA
## 17 Summer Healesville 2.8630137   6.46  NA                NA        NA
## 18 Summer Healesville 4.7260274   5.85  NA                NA        NA
## 19 Summer Healesville 3.7863014    4.8  NA                NA        NA
## 20 Spring Healesville 4.5232877   7.88  NA                NA        NA
## 21 Summer     Monarto 4.7726027   7.51  NA                NA        NA
## 22 Summer Healesville 2.7671233   5.82  NA                NA        NA
## 23 Summer     Monarto 4.5589041   8.59  NA                NA        NA
## 24 Winter Healesville 1.3424658   5.42  NA                NA        NA
## 25 Summer     Monarto 1.7917808   6.67  NA                NA        NA
## 26 Summer Healesville 0.7534247   2.72  NA                NA        NA
## 27 Winter     Monarto 1.5150685    4.6  NA                NA        NA
## 28 Winter Healesville 1.4000000   5.82  NA                NA        NA
## 29 Winter Healesville 1.2986301    5.1  NA                NA        NA
## 30 Summer     Monarto 4.5917808   9.82  NA                NA        NA
## 31 Winter Healesville 1.1123288   6.02  NA                NA        NA
## 32 Autumn Healesville 1.0164384    6.1  NA                NA        NA
## 33 Summer Healesville 1.4219178   6.62  NA                NA        NA
## 34 Summer Healesville 1.7315068   6.26  NA                NA        NA
## 35 Summer Healesville 1.7479452   7.42  NA                NA        NA
## 36 Winter     Monarto 1.6986301   8.75  NA                NA        NA
## 37 Winter Healesville 1.3041096  4.036  NA                NA        NA
## 38 Winter Healesville 1.3616438    6.4  NA                NA        NA
## 39 Winter     Monarto 1.5178082   8.15  NA                NA        NA
## 40 Winter Healesville 1.2630137   6.32  NA                NA        NA
## 41 Winter Healesville 1.2986301   5.76  NA                NA        NA
## 42 Winter Healesville 1.1753425   5.34  NA                NA        NA
## 43 Winter     Monarto 1.4657534   4.34  NA                NA        NA
## 44 Winter Healesville 1.3835616    N/A  NA                NA        NA
## 45 Winter Healesville 5.1041096   8.02  NA                NA        NA
## 46 Autumn Healesville 1.8246575   5.94  NA                NA        NA
## 47 Summer Healesville 2.7452055    N/A  NA                NA        NA
## 48 Spring Healesville 6.6273973   7.52  NA                NA        NA
## 49 Winter Healesville 1.1479452   6.54  NA                NA        NA
## 50 Summer Healesville 3.6904110   7.74  NA                NA        NA
## 51 Winter Healesville 4.3013699   9.16  NA                NA        NA
## 52 Spring Healesville 1.6767123   5.86  NA                NA        NA
##    UlcerationDepth Secondary.infection Total.tumour.volume  Barcode.sequence
## 1                                   NA                  NA GCTATCCT+AACAGGTG
## 2                                   NA                  NA ACGTTCAG+AGTTGTGC
## 3                                   NA                  NA CCGTAAGA+AGAGTCCA
## 4                                   NA                  NA CAACACCT+AGCTACCA
## 5                                   NA                  NA GATTACCG+TTGGACTG
## 6                                   NA                  NA CACTAGCT+TCGGATTC
## 7                                   NA                  NA AGTGGATC+TCGATGAC
## 8                                   NA                  NA ATCACACG+GTAGCGTA
## 9                                   NA                  NA AACGTCTG+AATGACGC
## 10                                  NA                  NA GTGCCATA+CTCCTAGT
## 11                                  NA                  NA ATCGCCAT+TGCGTAAC
## 12                                  NA                  NA GAACATCG+TCGACAAG
## 13                                  NA                  NA ACGACTTG+AACAGCGA
## 14                                  NA                  NA CCTTGTAG+CCTTGGAA
## 15                                  NA                  NA GAGCAGTA+TGAGCTGT
## 16                                  NA                  NA GATAGGCT+CAGTCACA
## 17                                  NA                  NA TGAAGACG+ACATGCCA
## 18                                  NA                  NA AACGTCTG+AATGACGC
## 19                                  NA                  NA GTCATCGA+GATCCACT
## 20                                  NA                  NA AGCGTGTT+CTTCGCAA
## 21                                  NA                  NA GCAAGATC+AGTCGAAG
## 22                                  NA                  NA TTGCGAAG+AACACGCT
## 23                                  NA                  NA TGCTTCCA+CGATCGAT
## 24                                  NA                  NA TGAGCTAG+GAACGGTT
## 25                                  NA                  NA CTTAGGAC+ACTCCTAC
## 26                                  NA                  NA TGGTACAG+TCGAGAGT
## 27                                  NA                  NA ACCACGAT+GTCTGCAA
## 28                                  NA                  NA ACTGAGGT+GTCAACAG
## 29                                  NA                  NA ACGACTTG+AACAGCGA
## 30                                  NA                  NA TCAGGCTT+ATCATGCG
## 31                                  NA                  NA ATGCCTGT+AGATTGCG
## 32                                  NA                  NA CCGTAAGA+AGAGTCCA
## 33                                  NA                  NA TTCCAAGG+CTACAAGG
## 34                                  NA                  NA GTAGGAGT+GTCCTAAG
## 35                                  NA                  NA CTTCGACT+GATCTTGC
## 36                                  NA                  NA TTGGACGT+GAAGTGCT
## 37                                  NA                  NA TATCGGTC+TACCGGAT
## 38                                  NA                  NA TACGCCTT+GCTACTCT
## 39                                  NA                  NA CGTGTGTA+CCAACGAA
## 40                                  NA                  NA CCTTGTAG+CCTTGGAA
## 41                                  NA                  NA CGCAATCT+ACAGGCAT
## 42                                  NA                  NA AGTGGATC+TCGATGAC
## 43                                  NA                  NA CTGTTGAC+ACCTCAGT
## 44                                  NA                  NA AACCGTTC+CTAGCTCA
## 45                                  NA                  NA TAACCGGT+ATCGTCTC
## 46                                  NA                  NA TTGCAGAC+ATCGTGGT
## 47                                  NA                  NA GAGACGAT+ACCGGTTA
## 48                                  NA                  NA CTTGTCGA+CGATGTTC
## 49                                  NA                  NA TACGCTAC+CGTGTGAT
## 50                                  NA                  NA GAATCCGA+AGCTAGTG
## 51                                  NA                  NA GCACAACT+CTGAACGT
## 52                                  NA                  NA TGTGACTG+AGCCTATC
##    PF.Clusters X..of.the.lane X..Perfect.barcode X..One.mismatch.barcode
## 1    5,338,890           1.00              98.08                    1.92
## 2   14,968,459           3.00              97.82                    2.18
## 3    4,340,498           1.00              97.95                    2.05
## 4   16,944,152           3.00              97.98                    2.02
## 5    5,125,566           1.00              96.21                    3.79
## 6    5,317,767           1.00              97.33                    2.67
## 7    2,690,632           1.00              97.77                    2.23
## 8   12,134,750           2.00              96.94                    3.06
## 9    3,809,607           1.00              96.73                    3.27
## 10   9,606,720           2.00              97.52                    2.48
## 11   6,018,884           1.00              97.88                    2.12
## 12   6,861,544           1.00              97.76                    2.24
## 13   3,000,501           1.00              97.96                    2.04
## 14   5,927,942           1.00              97.40                    2.60
## 15   1,702,608           0.00              98.07                    1.93
## 16   7,628,257           2.00              98.34                    1.66
## 17  13,381,058           3.00              96.98                    3.02
## 18   8,041,818           1.73              96.32                    3.68
## 19   5,758,111           1.24              96.72                    3.28
## 20   5,774,637           1.24              96.76                    3.24
## 21     348,621           0.08              97.24                    2.76
## 22   6,376,997           1.37              97.60                    2.40
## 23   9,181,668           1.98              98.08                    1.92
## 24   3,740,878           0.81              97.15                    2.85
## 25  11,133,566           2.40              97.85                    2.15
## 26   7,887,375           1.70              97.57                    2.43
## 27   7,308,907           1.58              97.70                    2.30
## 28   2,984,930           0.64              97.59                    2.41
## 29          98           0.00              94.90                    5.10
## 30   4,879,629           1.05              97.38                    2.62
## 31   4,857,395           1.05              97.58                    2.42
## 32   1,074,345           0.23              97.56                    2.44
## 33  10,669,237           2.30              97.64                    2.36
## 34  10,603,179           2.29              97.34                    2.66
## 35   5,642,025           1.22              96.37                    3.63
## 36  10,821,895           2.33              96.02                    3.98
## 37   8,693,920           1.87              98.13                    1.87
## 38   3,422,010           0.74              97.72                    2.28
## 39   4,811,453           1.04              97.67                    2.33
## 40   5,913,586           1.27              97.05                    2.95
## 41   8,306,537           1.79              96.94                    3.06
## 42   3,528,471           0.76              97.24                    2.76
## 43          23           0.00             100.00                     NaN
## 44   7,247,793           1.56              97.63                    2.37
## 45   6,908,416           1.49              97.09                    2.91
## 46   4,475,461           0.96              95.05                    4.95
## 47  12,024,421           2.59              95.37                    4.63
## 48  10,751,797           2.32              97.49                    2.51
## 49  13,015,081           2.81              97.36                    2.64
## 50   3,336,983           0.72              97.10                    2.90
## 51   6,826,313           1.47              97.66                    2.34
## 52   4,566,180           0.98              97.68                    2.32
##    Yield..Mbases. X..PF.Cluster X.....Q30 Mean.Quality X.2 X.3 X.4
## 1           1,612           100     93.69        35.91  NA  NA  NA
## 2           4,520           100     93.68        35.91  NA  NA  NA
## 3           1,311           100     93.54        35.88  NA  NA  NA
## 4           5,117           100     93.05        35.78  NA  NA  NA
## 5           1,548           100     93.89        35.94  NA  NA  NA
## 6           1,606           100     93.24        35.83  NA  NA  NA
## 7             813           100     93.77        35.93  NA  NA  NA
## 8           3,665           100     93.34        35.84  NA  NA  NA
## 9           1,151           100     93.07        35.79  NA  NA  NA
## 10          2,901           100     93.25        35.83  NA  NA  NA
## 11          1,818           100     93.58        35.89  NA  NA  NA
## 12          2,072           100     93.49        35.87  NA  NA  NA
## 13            906           100     93.70        35.91  NA  NA  NA
## 14          1,790           100     93.86        35.94  NA  NA  NA
## 15            514           100     93.58        35.89  NA  NA  NA
## 16          2,304           100     93.62        35.89  NA  NA  NA
## 17          4,041           100     93.36        35.84  NA  NA  NA
## 18          2,429           100     92.18        35.63  NA  NA  NA
## 19          1,739           100     91.87        35.55  NA  NA  NA
## 20          1,744           100     91.56        35.50  NA  NA  NA
## 21            105           100     92.77        35.73  NA  NA  NA
## 22          1,926           100     92.44        35.67  NA  NA  NA
## 23          2,773           100     93.04        35.79  NA  NA  NA
## 24          1,130           100     93.03        35.79  NA  NA  NA
## 25          3,362           100     92.47        35.68  NA  NA  NA
## 26          2,382           100     93.15        35.81  NA  NA  NA
## 27          2,207           100     92.84        35.75  NA  NA  NA
## 28            901           100     92.49        35.69  NA  NA  NA
## 29              0           100     89.11        34.91  NA  NA  NA
## 30          1,474           100     92.63        35.71  NA  NA  NA
## 31          1,467           100     92.64        35.72  NA  NA  NA
## 32            324           100     92.66        35.72  NA  NA  NA
## 33          3,222           100     92.79        35.74  NA  NA  NA
## 34          3,202           100     92.01        35.58  NA  NA  NA
## 35          1,704           100     91.87        35.56  NA  NA  NA
## 36          3,268           100     92.72        35.73  NA  NA  NA
## 37          2,626           100     92.32        35.65  NA  NA  NA
## 38          1,033           100     92.94        35.76  NA  NA  NA
## 39          1,453           100     92.30        35.65  NA  NA  NA
## 40          1,786           100     92.92        35.77  NA  NA  NA
## 41          2,509           100     91.86        35.55  NA  NA  NA
## 42          1,066           100     93.08        35.79  NA  NA  NA
## 43              0           100     95.85        36.37  NA  NA  NA
## 44          2,189           100     92.28        35.63  NA  NA  NA
## 45          2,086           100     91.90        35.56  NA  NA  NA
## 46          1,352           100     91.42        35.46  NA  NA  NA
## 47          3,631           100     93.06        35.79  NA  NA  NA
## 48          3,247           100     92.77        35.74  NA  NA  NA
## 49          3,931           100     92.61        35.70  NA  NA  NA
## 50          1,008           100     93.09        35.79  NA  NA  NA
## 51          2,062           100     92.67        35.71  NA  NA  NA
## 52          1,379           100     92.13        35.61  NA  NA  NA

ssCapS <- ssCapS[ssCapS$Sample %in% colnames(x),]

dim(ssCapS)

## [1] 48 31

x1 <- x[,which(colnames(x) %in% ssCapS$Sample)]
dim(x1)

## [1] 310  48

x1 <- x1[which(rowMeans(x1)>10),]
dim(x1)

## [1] 261  48

ssCapS %>%  kbl(caption="Sample sheet")%>% kable_styling("hover",full_width=FALSE) 

Sample sheet

	Lane	Project	Client.ID	Sample	X	Category	Sex.Age.Season	Sex	Age	Season	Location	Agebyyear	Weight	X.1	Number.of.Tumours	Ulcerated	UlcerationDepth	Secondary.infection	Total.tumour.volume	Barcode.sequence	PF.Clusters	X..of.the.lane	X..Perfect.barcode	X..One.mismatch.barcode	Yield..Mbases.	X..PF.Cluster	X…..Q30	Mean.Quality	X.2	X.3	X.4
1	1	default	S2_B99008	6102468	NA	Captive	FAU	F	A	Summer	Monarto	3.7369863	8.12	NA	NA	NA		NA	NA	GCTATCCT+AACAGGTG	5,338,890	1.00	98.08	1.92	1,612	100	93.69	35.91	NA	NA	NA
2	1	default	S2_Bonney	6102469	NA	Captive	FAU	F	A	Summer	Healesville	2.5863014	5.4	NA	NA	NA		NA	NA	ACGTTCAG+AGTTGTGC	14,968,459	3.00	97.82	2.18	4,520	100	93.68	35.91	NA	NA	NA
3	1	default	S3_Jessie	6102476	NA	Captive	FAW	F	A	Winter	Monarto	2.1945205	12.95	NA	NA	NA		NA	NA	CCGTAAGA+AGAGTCCA	4,340,498	1.00	97.95	2.05	1,311	100	93.54	35.88	NA	NA	NA
4	1	default	S3_McCauley	6102477	NA	Captive	MAA	M	A	Autumn	Healesville	3.0328767	8.52	NA	NA	NA		NA	NA	CAACACCT+AGCTACCA	16,944,152	3.00	97.98	2.02	5,117	100	93.05	35.78	NA	NA	NA
5	1	default	S5_Fenton	6102484	NA	Captive	MAU	M	A	Summer	Monarto	3.0000000	N/A	NA	NA	NA		NA	NA	GATTACCG+TTGGACTG	5,125,566	1.00	96.21	3.79	1,548	100	93.89	35.94	NA	NA	NA
6	1	default	S5_Bryan Adams	6102485	NA	Captive	MAW	M	A	Winter	Healesville	4.1616438	8.6	NA	NA	NA		NA	NA	CACTAGCT+TCGGATTC	5,317,767	1.00	97.33	2.67	1,606	100	93.24	35.83	NA	NA	NA
7	1	default	S7_Fredo	6102496	NA	Captive	MAA	M	A	Autumn	Healesville	2.9945205	8.8	NA	NA	NA		NA	NA	AGTGGATC+TCGATGAC	2,690,632	1.00	97.77	2.23	813	100	93.77	35.93	NA	NA	NA
8	1	default	S7_Piglet	6102497	NA	Captive	FSU	F	S	Summer	Healesville	1.5890411	4.76	NA	NA	NA		NA	NA	ATCACACG+GTAGCGTA	12,134,750	2.00	96.94	3.06	3,665	100	93.34	35.84	NA	NA	NA
9	1	default	S10_McMahon	6102514	NA	Captive	MSP	M	S	Spring	Healesville	1.6493151	8.06	NA	NA	NA		NA	NA	AACGTCTG+AATGACGC	3,809,607	1.00	96.73	3.27	1,151	100	93.07	35.79	NA	NA	NA
10	1	default	S10_Logan	6102515	NA	Captive	MAU	M	A	Summer	Healesville	5.6000000	9.02	NA	NA	NA		NA	NA	GTGCCATA+CTCCTAGT	9,606,720	2.00	97.52	2.48	2,901	100	93.25	35.83	NA	NA	NA
11	1	default	S11_Xuki	6102520	NA	Captive	MAA	M	A	Autumn	Monarto	2.0767123	7.95	NA	NA	NA		NA	NA	ATCGCCAT+TGCGTAAC	6,018,884	1.00	97.88	2.12	1,818	100	93.58	35.89	NA	NA	NA
12	1	default	S11_Kalamata	6102521	NA	Captive	MSW	M	S	Winter	Healesville	1.2849315	5.58	NA	NA	NA		NA	NA	GAACATCG+TCGACAAG	6,861,544	1.00	97.76	2.24	2,072	100	93.49	35.87	NA	NA	NA
13	1	default	S12_Wally	6102528	NA	Captive	MAA	M	A	Autumn	Monarto	2.1616438	N/A	NA	NA	NA		NA	NA	ACGACTTG+AACAGCGA	3,000,501	1.00	97.96	2.04	906	100	93.70	35.91	NA	NA	NA
14	1	default	S12_Naz	6102529	NA	Captive	MAW	M	A	Winter	Monarto	4.4547945	N/A	NA	NA	NA		NA	NA	CCTTGTAG+CCTTGGAA	5,927,942	1.00	97.40	2.60	1,790	100	93.86	35.94	NA	NA	NA
15	1	default	S13_Humbug	6102530	NA	Captive	MAU	M	A	Summer	Monarto	2.5397260	7.94	NA	NA	NA		NA	NA	GAGCAGTA+TGAGCTGT	1,702,608	0.00	98.07	1.93	514	100	93.58	35.89	NA	NA	NA
16	1	default	S13_Xerxes	6102535	NA	Captive	MAA	M	A	Autumn	Monarto	2.0767123	N/A	NA	NA	NA		NA	NA	GATAGGCT+CAGTCACA	7,628,257	2.00	98.34	1.66	2,304	100	93.62	35.89	NA	NA	NA
17	1	default	S14_Catani	6102538	NA	Captive	FAU	F	A	Summer	Healesville	2.8630137	6.46	NA	NA	NA		NA	NA	TGAAGACG+ACATGCCA	13,381,058	3.00	96.98	3.02	4,041	100	93.36	35.84	NA	NA	NA
18	2	default	S15_Amelia	6102541	NA	Captive	FAU	F	A	Summer	Healesville	4.7260274	5.85	NA	NA	NA		NA	NA	AACGTCTG+AATGACGC	8,041,818	1.73	96.32	3.68	2,429	100	92.18	35.63	NA	NA	NA
19	2	default	S15_London	6102542	NA	Captive	FAU	F	A	Summer	Healesville	3.7863014	4.8	NA	NA	NA		NA	NA	GTCATCGA+GATCCACT	5,758,111	1.24	96.72	3.28	1,739	100	91.87	35.55	NA	NA	NA
20	2	default	S15_Elyse	6102543	NA	Captive	FAP	F	A	Spring	Healesville	4.5232877	7.88	NA	NA	NA		NA	NA	AGCGTGTT+CTTCGCAA	5,774,637	1.24	96.76	3.24	1,744	100	91.56	35.50	NA	NA	NA
22	2	default	S16_Milla	6102548	NA	Captive	FAU	F	A	Summer	Healesville	2.7671233	5.82	NA	NA	NA		NA	NA	TTGCGAAG+AACACGCT	6,376,997	1.37	97.60	2.40	1,926	100	92.44	35.67	NA	NA	NA
23	2	default	S16_Maleficent	6102549	NA	Captive	FAU	F	A	Summer	Monarto	4.5589041	8.59	NA	NA	NA		NA	NA	TGCTTCCA+CGATCGAT	9,181,668	1.98	98.08	1.92	2,773	100	93.04	35.79	NA	NA	NA
24	2	default	S17_Anirak	6102553	NA	Captive	FSW	F	S	Winter	Healesville	1.3424658	5.42	NA	NA	NA		NA	NA	TGAGCTAG+GAACGGTT	3,740,878	0.81	97.15	2.85	1,130	100	93.03	35.79	NA	NA	NA
25	2	default	S17_Poly	6102554	NA	Captive	FSU	F	S	Summer	Monarto	1.7917808	6.67	NA	NA	NA		NA	NA	CTTAGGAC+ACTCCTAC	11,133,566	2.40	97.85	2.15	3,362	100	92.47	35.68	NA	NA	NA
26	2	default	S17_Methyl	6102556	NA	Captive	FSU	F	S	Summer	Healesville	0.7534247	2.72	NA	NA	NA		NA	NA	TGGTACAG+TCGAGAGT	7,887,375	1.70	97.57	2.43	2,382	100	93.15	35.81	NA	NA	NA
27	2	default	S18_Zara	6102558	NA	Captive	FSW	F	S	Winter	Monarto	1.5150685	4.6	NA	NA	NA		NA	NA	ACCACGAT+GTCTGCAA	7,308,907	1.58	97.70	2.30	2,207	100	92.84	35.75	NA	NA	NA
28	2	default	S18_Maze	6102559	NA	Captive	FSW	F	S	Winter	Healesville	1.4000000	5.82	NA	NA	NA		NA	NA	ACTGAGGT+GTCAACAG	2,984,930	0.64	97.59	2.41	901	100	92.49	35.69	NA	NA	NA
30	2	default	S19_Morocco	6102564	NA	Captive	MAU	M	A	Summer	Monarto	4.5917808	9.82	NA	NA	NA		NA	NA	TCAGGCTT+ATCATGCG	4,879,629	1.05	97.38	2.62	1,474	100	92.63	35.71	NA	NA	NA
31	2	default	S19_Pecorino	6102565	NA	Captive	FSW	F	S	Winter	Healesville	1.1123288	6.02	NA	NA	NA		NA	NA	ATGCCTGT+AGATTGCG	4,857,395	1.05	97.58	2.42	1,467	100	92.64	35.72	NA	NA	NA
33	2	default	S20_Muasy	6102569	NA	Captive	MSU	M	S	Summer	Healesville	1.4219178	6.62	NA	NA	NA		NA	NA	TTCCAAGG+CTACAAGG	10,669,237	2.30	97.64	2.36	3,222	100	92.79	35.74	NA	NA	NA
34	2	default	S20_Fuse	6102570	NA	Captive	MSU	M	S	Summer	Healesville	1.7315068	6.26	NA	NA	NA		NA	NA	GTAGGAGT+GTCCTAAG	10,603,179	2.29	97.34	2.66	3,202	100	92.01	35.58	NA	NA	NA
35	2	default	S20_Joseph	6102571	NA	Captive	MSU	M	S	Summer	Healesville	1.7479452	7.42	NA	NA	NA		NA	NA	CTTCGACT+GATCTTGC	5,642,025	1.22	96.37	3.63	1,704	100	91.87	35.56	NA	NA	NA
36	2	default	S21_Moama	6102576	NA	Captive	MAW	M	A	Winter	Monarto	1.6986301	8.75	NA	NA	NA		NA	NA	TTGGACGT+GAAGTGCT	10,821,895	2.33	96.02	3.98	3,268	100	92.72	35.73	NA	NA	NA
37	2	default	S21_Nauset	6102577	NA	Captive	MSW	M	S	Winter	Healesville	1.3041096	4.036	NA	NA	NA		NA	NA	TATCGGTC+TACCGGAT	8,693,920	1.87	98.13	1.87	2,626	100	92.32	35.65	NA	NA	NA
38	2	default	S21_Graham	6102578	NA	Captive	MSW	M	S	Winter	Healesville	1.3616438	6.4	NA	NA	NA		NA	NA	TACGCCTT+GCTACTCT	3,422,010	0.74	97.72	2.28	1,033	100	92.94	35.76	NA	NA	NA
39	2	default	S22_Yanni	6102581	NA	Captive	MSW	M	S	Winter	Monarto	1.5178082	8.15	NA	NA	NA		NA	NA	CGTGTGTA+CCAACGAA	4,811,453	1.04	97.67	2.33	1,453	100	92.30	35.65	NA	NA	NA
40	2	default	S22_Gherkin	6102582	NA	Captive	MSW	M	S	Winter	Healesville	1.2630137	6.32	NA	NA	NA		NA	NA	CCTTGTAG+CCTTGGAA	5,913,586	1.27	97.05	2.95	1,786	100	92.92	35.77	NA	NA	NA
41	2	default	S22_Stilton	6102584	NA	Captive	MSW	M	S	Winter	Healesville	1.2986301	5.76	NA	NA	NA		NA	NA	CGCAATCT+ACAGGCAT	8,306,537	1.79	96.94	3.06	2,509	100	91.86	35.55	NA	NA	NA
42	2	default	S23_Osiris	6102588	NA	Captive	MSW	M	S	Winter	Healesville	1.1753425	5.34	NA	NA	NA		NA	NA	AGTGGATC+TCGATGAC	3,528,471	0.76	97.24	2.76	1,066	100	93.08	35.79	NA	NA	NA
44	2	default	S23_Quince	6102590	NA	Captive	MSW	M	S	Winter	Healesville	1.3835616	N/A	NA	NA	NA		NA	NA	AACCGTTC+CTAGCTCA	7,247,793	1.56	97.63	2.37	2,189	100	92.28	35.63	NA	NA	NA
45	2	default	S24_Malfoy	6102595	NA	Captive	MAW	M	A	Winter	Healesville	5.1041096	8.02	NA	NA	NA		NA	NA	TAACCGGT+ATCGTCTC	6,908,416	1.49	97.09	2.91	2,086	100	91.90	35.56	NA	NA	NA
46	2	default	S24_Licorice	6102596	NA	Captive	FSA	F	S	Autumn	Healesville	1.8246575	5.94	NA	NA	NA		NA	NA	TTGCAGAC+ATCGTGGT	4,475,461	0.96	95.05	4.95	1,352	100	91.42	35.46	NA	NA	NA
47	2	default	S25_Tofu	6102599	NA	Captive	FAU	F	A	Summer	Healesville	2.7452055	N/A	NA	NA	NA		NA	NA	GAGACGAT+ACCGGTTA	12,024,421	2.59	95.37	4.63	3,631	100	93.06	35.79	NA	NA	NA
48	2	default	S25_Colette	6102600	NA	Captive	FAP	F	A	Spring	Healesville	6.6273973	7.52	NA	NA	NA		NA	NA	CTTGTCGA+CGATGTTC	10,751,797	2.32	97.49	2.51	3,247	100	92.77	35.74	NA	NA	NA
49	2	default	S25_CannonBall	6102601	NA	Captive	MSW	M	S	Winter	Healesville	1.1479452	6.54	NA	NA	NA		NA	NA	TACGCTAC+CGTGTGAT	13,015,081	2.81	97.36	2.64	3,931	100	92.61	35.70	NA	NA	NA
50	2	default	S25_Cornelius	6102602	NA	Captive	MAU	M	A	Summer	Healesville	3.6904110	7.74	NA	NA	NA		NA	NA	GAATCCGA+AGCTAGTG	3,336,983	0.72	97.10	2.90	1,008	100	93.09	35.79	NA	NA	NA
51	2	default	S25_Mardook	6102603	NA	Captive	MAW	M	A	Winter	Healesville	4.3013699	9.16	NA	NA	NA		NA	NA	GCACAACT+CTGAACGT	6,826,313	1.47	97.66	2.34	2,062	100	92.67	35.71	NA	NA	NA
52	2	default	S25_Negroni	6102604	NA	Captive	FSP	F	S	Spring	Healesville	1.6767123	5.86	NA	NA	NA		NA	NA	TGTGACTG+AGCCTATC	4,566,180	0.98	97.68	2.32	1,379	100	92.13	35.61	NA	NA	NA

col <- as.character(as.numeric(grepl("F",ssHS$Sex)))
col <- gsub("0","grey",col)
col <- gsub("1","black",col)

mds<-plotMDS(x1,pch=19,cex=3,col=col,main="MDS plot: Comparison of Male and Female Captive devils")
mtext("Male=Grey, Female=Black")

ssCapS$CapLoc <- grepl("Monarto",ssCapS$Location)
ssCapS$MvF <- factor(ssCapS$Sex)
ssCapS$SubAdvAd <- as.numeric(factor(ssCapS$Age,levels=c("S","A"),ordered = TRUE))
ssCapS$Season <- factor(ssCapS$Season)
str(ssCapS)

## 'data.frame':    48 obs. of  34 variables:
##  $ Lane                   : int  1 1 1 1 1 1 1 1 1 1 ...
##  $ Project                : chr  "default" "default" "default" "default" ...
##  $ Client.ID              : chr  "S2_B99008" "S2_Bonney" "S3_Jessie" "S3_McCauley" ...
##  $ Sample                 : int  6102468 6102469 6102476 6102477 6102484 6102485 6102496 6102497 6102514 6102515 ...
##  $ X                      : logi  NA NA NA NA NA NA ...
##  $ Category               : chr  "Captive" "Captive" "Captive" "Captive" ...
##  $ Sex.Age.Season         : chr  "FAU" "FAU" "FAW" "MAA" ...
##  $ Sex                    : chr  "F" "F" "F" "M" ...
##  $ Age                    : chr  "A" "A" "A" "A" ...
##  $ Season                 : Factor w/ 4 levels "Autumn","Spring",..: 3 3 4 1 3 4 1 3 2 3 ...
##  $ Location               : chr  "Monarto" "Healesville" "Monarto" "Healesville" ...
##  $ Agebyyear              : num  3.74 2.59 2.19 3.03 3 ...
##  $ Weight                 : chr  "8.12" "5.4" "12.95" "8.52" ...
##  $ X.1                    : logi  NA NA NA NA NA NA ...
##  $ Number.of.Tumours      : int  NA NA NA NA NA NA NA NA NA NA ...
##  $ Ulcerated              : logi  NA NA NA NA NA NA ...
##  $ UlcerationDepth        : chr  "" "" "" "" ...
##  $ Secondary.infection    : logi  NA NA NA NA NA NA ...
##  $ Total.tumour.volume    : num  NA NA NA NA NA NA NA NA NA NA ...
##  $ Barcode.sequence       : chr  "GCTATCCT+AACAGGTG" "ACGTTCAG+AGTTGTGC" "CCGTAAGA+AGAGTCCA" "CAACACCT+AGCTACCA" ...
##  $ PF.Clusters            : chr  "5,338,890" "14,968,459" "4,340,498" "16,944,152" ...
##  $ X..of.the.lane         : num  1 3 1 3 1 1 1 2 1 2 ...
##  $ X..Perfect.barcode     : num  98.1 97.8 98 98 96.2 ...
##  $ X..One.mismatch.barcode: num  1.92 2.18 2.05 2.02 3.79 2.67 2.23 3.06 3.27 2.48 ...
##  $ Yield..Mbases.         : chr  "1,612" "4,520" "1,311" "5,117" ...
##  $ X..PF.Cluster          : int  100 100 100 100 100 100 100 100 100 100 ...
##  $ X.....Q30              : num  93.7 93.7 93.5 93 93.9 ...
##  $ Mean.Quality           : num  35.9 35.9 35.9 35.8 35.9 ...
##  $ X.2                    : logi  NA NA NA NA NA NA ...
##  $ X.3                    : logi  NA NA NA NA NA NA ...
##  $ X.4                    : logi  NA NA NA NA NA NA ...
##  $ CapLoc                 : logi  TRUE FALSE TRUE FALSE TRUE FALSE ...
##  $ MvF                    : Factor w/ 2 levels "F","M": 1 1 1 2 2 2 2 1 2 2 ...
##  $ SubAdvAd               : num  2 2 2 2 2 2 2 1 1 2 ...

dds <- DESeqDataSetFromMatrix(countData = x1 , colData = ssCapS, design = ~ Season +  SubAdvAd + CapLoc + MvF)

## converting counts to integer mode

##   the design formula contains one or more numeric variables with integer values,
##   specifying a model with increasing fold change for higher values.
##   did you mean for this to be a factor? if so, first convert
##   this variable to a factor using the factor() function

res <- DESeq(dds)

## estimating size factors

## estimating dispersions

## gene-wise dispersion estimates

## mean-dispersion relationship

## final dispersion estimates

## fitting model and testing

## -- replacing outliers and refitting for 10 genes
## -- DESeq argument 'minReplicatesForReplace' = 7 
## -- original counts are preserved in counts(dds)

## estimating dispersions

## fitting model and testing

z<- results(res)
vsd <- vst(dds, blind=FALSE,nsub=nrow(x1)-50)
zz<-cbind(as.data.frame(z),assay(vsd))
dge<-as.data.frame(zz[order(zz$pvalue),])
dgeCapS <- dge
sig <- subset(dge,padj<0.05)
sig1_up <- rownames(subset(sig,log2FoldChange>0))
sig1_dn <- rownames(subset(sig,log2FoldChange<0))
length(sig1_up)

## [1] 0

maplot(dgeCapS,"Male vs Female Captive devils")

make_volcano(dgeCapS,"Male vs Female Captive devils")

col <- as.character(as.numeric(grepl("F",ssCapS$Sex)))
col <- gsub("0","grey",col)
col <- gsub("1","black",col)
mtext("Sub-Adult=Orange, Adult=Brown")

sig[1:100,1:6] %>%
  kbl(caption="Comparison of Male vs Female Captive devils") %>%
  kable_paper("hover", full_width = F)

Comparison of Male vs Female Captive devils

	baseMean	log2FoldChange	lfcSE	stat	pvalue	padj
IDO1	5484.3259	-2.890992	0.6674776	-4.331220	0.0000148	0.0037961
SIRPA	215.9456	-1.339309	0.3773424	-3.549321	0.0003862	0.0446010
IL18	1718.3609	-1.258023	0.3626617	-3.468862	0.0005227	0.0446010
NA	NA	NA	NA	NA	NA	NA
NA.1	NA	NA	NA	NA	NA	NA
NA.2	NA	NA	NA	NA	NA	NA
NA.3	NA	NA	NA	NA	NA	NA
NA.4	NA	NA	NA	NA	NA	NA
NA.5	NA	NA	NA	NA	NA	NA
NA.6	NA	NA	NA	NA	NA	NA
NA.7	NA	NA	NA	NA	NA	NA
NA.8	NA	NA	NA	NA	NA	NA
NA.9	NA	NA	NA	NA	NA	NA
NA.10	NA	NA	NA	NA	NA	NA
NA.11	NA	NA	NA	NA	NA	NA
NA.12	NA	NA	NA	NA	NA	NA
NA.13	NA	NA	NA	NA	NA	NA
NA.14	NA	NA	NA	NA	NA	NA
NA.15	NA	NA	NA	NA	NA	NA
NA.16	NA	NA	NA	NA	NA	NA
NA.17	NA	NA	NA	NA	NA	NA
NA.18	NA	NA	NA	NA	NA	NA
NA.19	NA	NA	NA	NA	NA	NA
NA.20	NA	NA	NA	NA	NA	NA
NA.21	NA	NA	NA	NA	NA	NA
NA.22	NA	NA	NA	NA	NA	NA
NA.23	NA	NA	NA	NA	NA	NA
NA.24	NA	NA	NA	NA	NA	NA
NA.25	NA	NA	NA	NA	NA	NA
NA.26	NA	NA	NA	NA	NA	NA
NA.27	NA	NA	NA	NA	NA	NA
NA.28	NA	NA	NA	NA	NA	NA
NA.29	NA	NA	NA	NA	NA	NA
NA.30	NA	NA	NA	NA	NA	NA
NA.31	NA	NA	NA	NA	NA	NA
NA.32	NA	NA	NA	NA	NA	NA
NA.33	NA	NA	NA	NA	NA	NA
NA.34	NA	NA	NA	NA	NA	NA
NA.35	NA	NA	NA	NA	NA	NA
NA.36	NA	NA	NA	NA	NA	NA
NA.37	NA	NA	NA	NA	NA	NA
NA.38	NA	NA	NA	NA	NA	NA
NA.39	NA	NA	NA	NA	NA	NA
NA.40	NA	NA	NA	NA	NA	NA
NA.41	NA	NA	NA	NA	NA	NA
NA.42	NA	NA	NA	NA	NA	NA
NA.43	NA	NA	NA	NA	NA	NA
NA.44	NA	NA	NA	NA	NA	NA
NA.45	NA	NA	NA	NA	NA	NA
NA.46	NA	NA	NA	NA	NA	NA
NA.47	NA	NA	NA	NA	NA	NA
NA.48	NA	NA	NA	NA	NA	NA
NA.49	NA	NA	NA	NA	NA	NA
NA.50	NA	NA	NA	NA	NA	NA
NA.51	NA	NA	NA	NA	NA	NA
NA.52	NA	NA	NA	NA	NA	NA
NA.53	NA	NA	NA	NA	NA	NA
NA.54	NA	NA	NA	NA	NA	NA
NA.55	NA	NA	NA	NA	NA	NA
NA.56	NA	NA	NA	NA	NA	NA
NA.57	NA	NA	NA	NA	NA	NA
NA.58	NA	NA	NA	NA	NA	NA
NA.59	NA	NA	NA	NA	NA	NA
NA.60	NA	NA	NA	NA	NA	NA
NA.61	NA	NA	NA	NA	NA	NA
NA.62	NA	NA	NA	NA	NA	NA
NA.63	NA	NA	NA	NA	NA	NA
NA.64	NA	NA	NA	NA	NA	NA
NA.65	NA	NA	NA	NA	NA	NA
NA.66	NA	NA	NA	NA	NA	NA
NA.67	NA	NA	NA	NA	NA	NA
NA.68	NA	NA	NA	NA	NA	NA
NA.69	NA	NA	NA	NA	NA	NA
NA.70	NA	NA	NA	NA	NA	NA
NA.71	NA	NA	NA	NA	NA	NA
NA.72	NA	NA	NA	NA	NA	NA
NA.73	NA	NA	NA	NA	NA	NA
NA.74	NA	NA	NA	NA	NA	NA
NA.75	NA	NA	NA	NA	NA	NA
NA.76	NA	NA	NA	NA	NA	NA
NA.77	NA	NA	NA	NA	NA	NA
NA.78	NA	NA	NA	NA	NA	NA
NA.79	NA	NA	NA	NA	NA	NA
NA.80	NA	NA	NA	NA	NA	NA
NA.81	NA	NA	NA	NA	NA	NA
NA.82	NA	NA	NA	NA	NA	NA
NA.83	NA	NA	NA	NA	NA	NA
NA.84	NA	NA	NA	NA	NA	NA
NA.85	NA	NA	NA	NA	NA	NA
NA.86	NA	NA	NA	NA	NA	NA
NA.87	NA	NA	NA	NA	NA	NA
NA.88	NA	NA	NA	NA	NA	NA
NA.89	NA	NA	NA	NA	NA	NA
NA.90	NA	NA	NA	NA	NA	NA
NA.91	NA	NA	NA	NA	NA	NA
NA.92	NA	NA	NA	NA	NA	NA
NA.93	NA	NA	NA	NA	NA	NA
NA.94	NA	NA	NA	NA	NA	NA
NA.95	NA	NA	NA	NA	NA	NA
NA.96	NA	NA	NA	NA	NA	NA

write.table(dge,file="dgeCapS.tsv",sep="\t")

mx <- sig[,7:ncol(sig)]
mx <- head(mx,30)
colfunc <- colorRampPalette(c("blue", "white", "red"))
heatmap.2(as.matrix(mx),trace="none",scale="row",
          col=colfunc(25),ColSideColors=col,mar=c(5,12))