A Novel Computational Method for the Identification and Functional Analysis of Clinically Relevant Alternative Splicing Events in Cancer Genomic Data

Abstract

Alternative splicing – the process by which different exons in immature mRNA transcripts are excised and combined together to form a diverse range of final mRNAs – has been implicated in many different cancers, including breast cancer. Splice isoforms of various genes have been shown to contribute to more aggressive cancer phenotypes, and studying these variants in order to gain insights into potential novel targets of cancer therapy is an ongoing field of research. In this thesis, I describe a novel computational method designed to detect alternative exons based on full gene expression normalized individual exon expression using RNAseq data from TCGA patient samples. Furthermore, normalized exon expression between different molecular subtypes of breast cancer is compared to identify clinically relevant instances of alternative splicing in breast cancer. To validate our method, I also describe our analysis of the gene BIRC5, which exhibits splice isoforms that have been well documented. I not only demonstrate that our method successfully identified BIRC5 splicing events in the raw data, but also show that we can glean functional insights from our analysis. Following the successful validation of our method using BIRC5, I also introduce our analysis of the genes TCAIM and GCNT2, which were identified as candidate genes by our method. I demonstrate that our method has uncovered interesting, previously undescribed putative functional roles for these genes, which may lead to the identification of novel druggable targets in the fight against cancer.