A shiny application for pancan survival analysis with paralog/miRNA pairs and in vitro validation of miRNA synergism in liver cancer

Abstract

Emerging cancer survival tools can predict risk of disease and identify prognostic biomarkers. Multivariable Cox proportional hazards models with mRNA and microRNAs (miRNAs) expression can differentiate survival outcomes. Previous studies showed that genes that belong to the same pathways/families may act independently, and in a cancer-specific manner. In this thesis, cancer-dependent hazard ratios of paralog genes and sense-antisense strands of miRNAs were tested for TCGA PANCAN. The results were presented in a R/Shiny web application that provides gene-by-survival networks. The gene-by-survival network approach also was applied to the plasma membrane-endoplasmic reticulum (PM-ER) calcium channel geneset. Among paralogs, cancer-specific prognostic signatures and functional compartmentalization were observed. Some cancers like UVM, MESO emerged as hub cancers for PM-ER signalling. Further the proposed gene-by-survival network approach has been extended for miRNA-mRNA triplets that may act in synergy in hepatocellular carcinoma (HCC). Next, the effects of synergistic miRNA pairs provided by miRCoop algorithm were tested on cell viability and target gene expression for selected triplets. The results have revealed that the combinatorial miRNA treatments show promising results as RNAi therapeutics yet future studies with different doses and triplets are needed.