Abstract: Objective To identify genes associated with prognosis or differentiated type in gastric cancer from frequently mutated genes or highly-expressed genes, using large-scale genomic data from The Cancer Genome Atlas. Methods The somatic mutation data, RNAseqV2 data and clinical information were downloaded from the TCGA website. The frequency of deleterious somatic mutations for each gene was counted to select the frequently mutated genes. DESeq2 was used to analyze the gene expression data. Then survival analysis was performed on genes highly-expressed in the tumor tissue. Kaplan-Meier curves were generated by R-survival package, and significance was evaluated by log-rank test. Results The frequency for pathogenic mutations in PIK3CA and APC was significantly discordant between different grades of gastric cancer. 2 040 genes were up-regulated in tumor tissue, while 2 357 genes were down -regulated. Among the up-regulated genes, 7 genes were associated with poor prognosis of gastric cancer and one was associated with better prognosis. Conclusion In this study, by analyzing TCGA gastric cancer dataset, we have identified several genes associated with differentiation types or prognosis in gastric cancer. Our work will provide clues for future research on potential prognostic markers in clinical treatment.

Key words: Key words: TCGA gastric cancer dataset, somatic mutation, gene expression, prognosis