Pan-cancer Analysis Of Clinical Relevance Of Alternative Splicing Events In Human Cancers

Purpose: Alternative splicing represents a critical post-transcriptional regulation of gene expression,which contributes to the protein complexity and m RNA processing.Defects of alternative splicing including genetic alteration and/or altered expression of both pre-m RNA and trans-acting factors give rise to many cancers.By integrally analyzing clinical data and splicing data from TCGA and Splice Seq databases,a number of splicing events were found clinically relevant in tumor samples.Methods: Differential alternative splicing events in tumor tissues were screened by subtracting the mean PSI from tumor tissues and adjacent normal tissues,and the significance of the differences was assessed using the paired samples Student's test.Univariate cox regression analysis was used to assess the association of alternative splicing events with survival of cancer patients.K-mean cluster analysis was used to explore the correlation and difference at molecular level between different tumors.Independent prognostic splicing events were screened using a random forest survival analysis algorithm,and prognosis predictors were constructed and evaluated in each tumor type.Results: Alternative splicing of KLK2?KLK2₅1239?was found as a potential inducement of nonsense-mediated m RNA decay and associated with poor survival in prostate cancer.K-means clustering analysis indicated that alternative splicing events could be potentially used for molecular subtype classification of cancers.By random forest survival algorithm,prognostic prediction signatures with well performances were constructed for 31 cancers by using survival associated alternative splicing events.Furthermore,an online tool for visualization of Kaplan-Meier plots and different splicing plots of splicing events in 31 cancers was explored.Conclusion: Briefly,alternative splicing was found of significant clinical relevance with cancers.