Abstract

   Toxygates was originally released as a user-friendly interface to
   enhance the accessibility of the large-scale toxicogenomics database,
   Open TG-GATEs, generated by the Japanese Toxicogenomics Project. Since
   the original release, significant new functionality has been added to
   enable users to perform sophisticated computational analysis with only
   modest bioinformatics skills. The new features include an orthologous
   mode for data comparison among different species, interactive
   clustering and heatmap visualisation, enrichment analysis of gene sets,
   and user data uploading. In a case study, we use these new functions to
   study the hepatotoxicity of peroxisome proliferator-activated receptor
   alpha (PPARα) agonist WY-14643. Our findings suggest that WY-14643
   caused hypertrophy in the bile duct by intracellular Ca^2+
   dysregulation, which resulted in the induction of genes in a
   non-canonical WNT/Ca^2+ signalling pathway. With this new release of
   Toxygates, we provide a suite of tools that allow anyone to carry out
   in-depth analysis of toxicogenomics in Open TG-GATEs, and of any other
   dataset that is uploaded.

Introduction

   Biomedical research produces ever more data, which are often made
   publicly available at an early stage of investigation. The increasing
   popularity of open-access journals is now being matched by the
   availability of open data. In fact, some publication venues, for
   example GigaScience^[32]1, insist on publication of the data itself
   before any associated results may be published. This trend helps the
   research process, as important discoveries about a particular dataset
   can often be made by people who did not originally produce the data.

   One example of this trend is the well-studied database Open
   TG-GATEs^[33]2, a large transcriptome (as well as associated phenotypes
   and biological data) database produced by Japanese Toxicogenomics
   Project (TGP) during 2002–2006 for TGP1^[34]3 and 2007–2011 for TGP2.
   The data were collected by a collaboration of the National Institute of
   Biomedical Innovation (NIBIO), the National Institute of Health
   Sciences and 15 pharmaceutical companies. This project aimed to create
   an infrastructure to evaluate the safety/toxicity of compounds,
   including drug candidates, by using these collected data, an approach
   referred to as toxicogenomics. It is expected that the public databases
   in toxicogenomics will support assessment of the safety of drug
   candidates in the early stage of the drug discovery process. This
   dataset has now been independently investigated by several different
   groups^[35]4, [36]5 as well as internally at NIBIOHN (formerly NIBIO),
   as one of the largest databases in toxicogenomics. It consists of
   approximately 24,000 microarray samples of about 200 different
   compounds, studied in rat tissues in vivo and rat- or human-derived
   primary cultured hepatocytes in vitro. Both single and repeat dose
   samples are available. Experimental conditions are well-defined and
   documented, and in theory, this dataset is well suited to the
   investigation of mechanisms of action in drugs and toxins that have not
   yet been explained.

   However, data availability by itself is not always sufficient. In
   practice, even when data have been produced and published, considerable
   effort may be needed on behalf of interested third parties who wish to
   study them. When Open TG-GATEs had originally been released, it was
   necessary to perform substantial pre-processing before any
   investigation could commence. To lower the barrier to entry for new
   investigators, we developed Toxygates^[37]6
   ([38]http://toxygates.nibiohn.go.jp), the first version of which was
   made publicly available in 2013. At that time, the main achievement was
   to allow anybody who was interested to select samples quickly (e.g.,
   compounds, exposure times) and look at expression data, visualise time
   or dose series, and show their annotations such as Gene Ontology (GO)
   terms^[39]7 or KEGG pathways^[40]8. Even with this basic feature set,
   Toxygates has seen regular use by researchers from a wide variety of
   countries and institutions since its release. During the year 2016, a
   total of 194 unique external users from 16 countries accessed Toxygates
   17165 times to look at expression data. At the same time, this original
   version did not allow for sophisticated analysis and for further
   comprehensive analysis, such as clustering and enrichment, it was
   necessary to download the data and use external software. Analytical
   functions are important in practice to experimental biologists and
   thus, for user convenience, it is ideal to analyse data without
   downloading.

   In spite of the growing demand for analytical applications to lower
   this barrier to large-scale databases such as Open TG-GATEs, the number
   of such applications is still limited. To the best of our knowledge,
   there are only two public web applications that utilise Open TG-GATEs
   even today; LTMap and ToxDBScan. LTMap (http://tcm.zju.edu.cn/ltmap/)
   is a web tool to compare input gene lists with reference gene lists,
   and it utilises Open TG-GATEs for reference gene lists to output a
   ranked list of drugs generated by rank-based pattern-matching
   algorithm^[41]9. ToxDBScan is also a web-based application for
   similarity search, utilising both Open TG-GATEs and DrugMatrix as
   reference data^[42]10. This tool calculates the similarity scores based
   on the extended connectivity fingerprints (ECFP) to compare input gene
   lists and references, and performs pathway enrichment analysis of input