Construction of prognostic predictor by comprehensive analyzing alternative splicing events for Colon adenocarcinoma
Abstract Background Colon adenocarcinoma (COAD) is one of the most common malignant tumors with high incidence and mortality rates worldwide. Reliable prognostic biomarkers are needed to identify to guide clinical practice.Materials and Method Comprehensive gene expression with AS profiles for each patient were downloaded using SpliceSeq database from The Cancer Genome Atlas (TCGA). Cox regression analysis was conducted to screen prognostic AS events. R package “limma” was used to screen different expression genes (DEGs) between normal and tumor samples in COAD cohort. Venn plot analysis was performed between DEGs and prognostic AS events, the DEGs with prognostic AS events occurred(DEGAS) were obtained. Top 30 DEGAS in protein-protein interaction(PPI) analysis were bring in Cox proportional hazards regression to establish prognostic models.Results Totally, 350 patients were included in study. 22,451 AS events were detected. 2,004 AS events form 1439 genes showed as survival-associated AS events significantly. By overlapping the 1439 genes and 6455 DEGs, 211 DEGs with AS events were obtained. After construction of PPI, top 30 hub genes were included in multivariable analysis. Finally, a risk score based on 12 genes which associated with overall survival(OS) was established (P < 0.05). The area under the curve(AUC) is 0.782. And the risk score shows as an independent predictor (P < 0.001).Conclusions By exploring survival-associated AS events, a powerful prognostic predictor was built. We aim at proposing a novel method to provide treatment targets for COAD and guide clinical practice in the future.