The Platform For Gene Set Cancer Analysis And Its Applications On Histone Methylation Genes And Immune Checkpoint Genes

With the generation of big data on cancer genomes,accurate analysis and interpretation of massive omics data has become a new challenge.Moreover,the occurrence and development of cancer are usually the results of mutation and abnormal expression of a group of genes or pathways,because abnormal signals of a single gene may be covered by background noise.Thus,performing association analysis with big data of cancer multi-omics to study the molecular behavior of a group of genes in a variety of cancers from the perspective of large multi-omics samples is imperative,very useful and extremely urgent for cancer research.At present,multiple researches that integrate public data to build multi-omics analysis platforms for single gene,but a multi-omics analysis platform based on gene sets is still lacking.Using such a gene set-based analysis platform,we can analyze different functional gene sets in different cancer types,screen out key genes,and dig molecular mechanisms.Based on this concept,we have achieved the following three aspects.1.GSCALite: a web server for gene set cancer analysisWe developed an interactive web-based application named GSCALite for Gene Set Cancer Analysis to analyze and visualize the expression/variation/correlation of a gene set in cancers with flexible manner.GSCALite including following functional modules:(i)Differential expression in tumor vs normal,and the survival analysis;(ii)Genomic variations and their survival analysis;(iii)Gene expression associated cancer pathway activity;(iv)miRNA regulatory network for genes;(v)Drug sensitivity for genes;(vi)Normal tissue expression and e QTL for genes.GSCALite is a user-friendly web server for dynamic analysis and visualization of gene set in cancer and drug sensitivity correlation,which will be of broad utilities to cancer researchers.Availability: GSCALite is available on http://bioinfo.life.hust.edu.cn/web/GSCALite/.2.Histone methylation genes CBX2/EZH2 centered signaling pathway promotes the growth and metastasis of lung adenocarcinomaThrough integrating LUAD multi-omics data from GSCALite,we studied the mechanism of gene expression regulation of histone methylation genes(HMGs)at the levels of transcriptome,genomics,and epigenetics.The differential expression,correlation,survival and regulatory network analyses were applied.We found that CBX2 and EZH2 were up-regulated in LUAD much more than other HMGs,and synergistically affect the progression and prognosis of LUAD,promote cell proliferation and invasion,and inhibit apoptosis of A549 and H1299 cells.CBX2 depletion inhibits LUAD growth and metastasis in mice.The PPAR signaling pathway significantly inactivated in LUAD,which turns out to be the transcriptional target of CBX2 and EZH2.The inhibition of PPAR signaling pathway promotes cell proliferation and invasion.The upregulations of CBX2 and EZH2 were achieved by transcription factors(E2F1 and SOX4)and miRNAs(miR-30d-5p,miR-101-3p and let-7c-5p).Moreover,CBX2,EZH2,and their related genes were potentially correlated with the sensitivity to drugs inhibiting cell proliferation.Our study discovers that the regulatory axis centered by CBX2/EZH2 promotes the progression of LUAD,providing potential targets for LUAD diagnosis and therapy.3.Expression profile of immune checkpoints genes and its role in predicting immune therapy responseThrough comprehensive pan-tissue and pan-cancer survey of TCGA and FANTOM5 RNA-seq data,we classified immune checkpoint genes(ICGs)into three patterns according to their expression in tumor/immune cells,including immune cells dominated ICGs(IC-ICGs),tumor cells dominated ICGs(TC-ICGs)and tumor & immune cells bivariate ICGs(TIC-ICGs).The expression of IC-ICGs and TC-ICGs showed inverse trend in clinical relevance(high expression of IC-ICGs served as predictor of good prognosis,high expression of TC-ICGs served as bad in some cancers).The expression profile of ICGs showed that upregulated ICGs significantly positively correlates with tumor immune cell infiltration and promote better prognosis,and could define the immune status of tumor.Constructed ICG expression features and screened which associated with immune profile(immune cells infiltrations,cytotoxic T lymphocytes,tumor mutation burden)consistently across different cancer types.Further filtered out features by hill climbing algorithm to build a 5 ICGs-expression-features based model named ICGe to predict ICB response,and validated it in 6 independent datasets with different treatments and tissues,with AUC of 0.64-0.82.ICGe outperformed when compared with other m RNA based predictors.These results provide novel clues to ICB response and better understand of the mechanism of ICGs in ICB therapy.This research built a multi-omics analysis platform for cancer gene sets,which provided an important platform for the interpretation of public multi-omics data in cancer research.It is an important tool for experimentalists to easily access public data sets to validate candidate genes.At the same time,by applying the data and analysis process of the platform,we analyzed HMGs in LUAD and screened a regulatory axis centered by CBX2 and EZH2,which promotes the progress and metastasis of LUAD.In addition,the m RNA expression level of ICGs was applied to reveal the expression profile of ICGs in pan-cancer and its relationship with immune infiltration,patient survival,and immunotherapy response.