Abstract In recent years, the incidence and mortality of cervical cancer have increased worldwide. At the same time, increasing data have confirmed that miRNA-mRNA plays a positive or negative regulatory role in many cancers. This study attempted to screen effective miRNA-mRNA in the progression of cervical cancer, and to study the mechanism of miRNA-mRNA in the progression of cervical cancer. The expression profile data of [39]GSE7410, GSE 63514, GSE 86100 and TCGA-CESC were downloaded, and 34 overlapping differentially expressed genes (22 up-regulated and 12 down-regulated) and 166 miRNAs (74 down-regulated and 92 up-regulated) were screened through limma package. Then, miR-197-3p/TYMS pairs were obtained by PPI, functional enrichment, Kaplan-Meier plotter analysis, Cox univariate and multivariate analysis, risk modeling, WGCNA, qPCR and dual-luciferase experiments. The results showed that TYMS was an independent prognostic factor of cervical cancer, and its expression level was negatively correlated with cervical cancer tissue grade (TMN), tumor grade, age, microsatellite stability and tumor mutation load, and positively correlated with methyl expression in DNMT1, DNMT2, DNMT3A and DNMT3B. Functional experiments showed that TYMS knockout could promote the proliferation, migration and invasion of HeLa cells and reduce apoptosis. Overexpression of TYMS showed the opposite trend, miR-197-3p was negatively correlated with the expression of TYMS. MiR-197-3p inhibitor reversed the effect of si-TYMS on the proliferation of HeLa cells. In conclusion, these results reveal that TYMS plays a very important role in the prognosis and progression of cervical cancer, and has the potential to be thought of as cervical cancer biomarkers. At the same time, miR-197-3p/TYMS axis can regulate the deterioration of cervical cancer cells, which lays a foundation for the molecular diagnosis and treatment of cervical cancer. Keywords: cervical cancer, miR-197-3p/TYMS, proliferation, apoptosis, invasion and migration Introduction Cervical cancer is the most common female malignancy worldwide, and directly causes high incidence and mortality rates in women ([40]Wang et al.,2020). According to the database of the International Agency for Research on Cancer, there were more than 500,000 new cases of cervical cancer and 311,000 deaths in 2018 ([41]Stelzle et al.,2021). To date, surgery and radiotherapy are still the main treatment methods for cervical cancer, but recurrence, metastasis and drug resistance often occur after treatment ([42]Sharma et al.,2020; [43]Shin et al.,2020). The mechanism of cervical cancer is complex. Different genes, RNAs and signaling pathways are related to the tumorigenicity of cervical cancer ([44]Huang et al., 2020; [45]Rasmi and Sakthivel, 2020). Therefore, it is worthwhile to find new methods to study the basic mechanism of cervical cancer to improve treatment. RNA plays an important role in the regulation of several major biological processes affecting tumorigenesis and progression, and has always been at the forefront of tumor molecular mechanism research ([46]Goodall and Wickramasinghe, 2021; [47]Barbieri & Kouzarides, 2020). The combination of high-throughput technology and bioinformatics analysis can provide researchers with valuable data available in the form of public datasets to search for biomarkers and therapeutic targets ([48]Liu et al.,2020; [49]Chen et al., 2008). Gene Expression Omnibus (GEO) facilitates the submission, storage, and retrieval of heterogeneous datasets from high-throughput gene expression and genomic experiments ([50]Clough and Barrett, 2016; [51]Yang et al., 2020). The Cancer Genome Atlas (TCGA) is a comprehensive dataset that provides a unified data analysis pipeline for further exploration of oncogene signaling changes and their associated significance in cancer patient outcomes ([52]Hutter and Zenklusen, 2018; [53]Linehan and Ricketts, 2019). Therefore, the combination of GEO and TCGA data sets may provide an important perspective for the study of new biomarkers. There have been many reports on screening tumor biomarkers based on GEO and TCGA data, and a series of markers with high specificity and sensitivity have been found ([54]Ren et al., 2020; [55]Xu et al.,2020). Compared with traditional screening methods, bioinformation-based analysis of high-throughput data enables researchers to obtain stable and reliable biomarkers in a large number of clinical samples. In this study, we combined cervical cancer in GEO database and TCGA databases to screen differentially expressed genes and differentially expressed miRNAs. Based on the differentially expressed genes, the candidate gene TYMS was screened by functional annotation, pathway analysis, protein interaction (PPI) network construction, prognosis analysis, risk assessment model and WGCNA. Then, potential candidate pairs (miR-197-3p/TYMS) were screened by dual-luciferase experiment and qPCR. Subsequently, the effects of TYMS and miR-197-3p/TYMS on the progression of cervical cancer were analyzed by gain-of-function analysis. We hope to provide more extensive data support for the clinical application of TYMS and the miR-197-3p/TYMS axis in cervical cancer. Materials and Methods Data Collection and Preprocessing The original datasets were downloaded from geo database to compare the mRNA expression and miRNA expression between cervical cancer and normal tissues. The expression profile data of [56]GSE7410, [57]GSE63514, [58]GSE86100 and [59]GSE9750 were downloaded based on different platforms ([60]Table 1). The mRNA expression data of cervical cancer were downloaded from TCGA database for preprocessing. Clinical samples related to cervical cancer were selected. The data set includes 307 cervical cancer samples, 307 normal samples and corresponding clinical data. Then, Sangerbox tool 2.0 ([61]http://sangerbox.com/Index) corrects and normalizes the original expression data background of each GEO dataset, and eliminates batch effects and other irrelevant variables. TABLE 1. Information of GEO datasets. Dataset Platform Tumor Normal References