Dissecting Protein-RNA Interaction Network in Human Genome

Abstract

In eukaryotes, gene regulation is a complex multilevel process comprising of transcriptional, post-transcriptional, and post-translational control. Although the regulation at transcriptional and post-translational levels is gradually being understood, protein machinery and the mechanisms underlying the post-transcriptional regulation remain to be elucidated. In the first study of this dissertation, I designed and implemented a database of RNA Binding Protein (RBP) Expression and Disease Dynamics (READ-DB: darwin.soic.iupui.edu), a non-redundant, curated database of human RBPs. This RBP knowledge base includes data from different experimental studies providing a one stop portal for understanding the expression, evolutionary trajectories, and disease dynamics of RBPs in the context of post-transcriptional regulatory networks. Despite the existence of several experimental procedures to understand the function of RBPs, a lack of a proper computational method to profile differential occupancy limits the scope of research. In the second study, I built a scalable framework for comparing genome-wide protein occupancy profiles among cell-types data, to uncover alterations in protein-RNA interactomes. diffHunter (github.com/Sasanh/diffHunter), is a window based peak calling and profile comparison method that can efficiently store the base-pair level read information of every given sample in a NoSQL (Not Only SQL) database. It identifies and quantitates the genome-wide binding differences between a pair of samples in two stages: Peak Calling and Differential Binding Identification. Identifying such regions enables us to compare the biologically important regions that differ between two conditions. Finally, I studied A-to- I RNA editing as one of the special functions of an RBPs’ family. ADAR family RBPs are the primary driver in the conversion of adenosine to inosine (A-to-I) within mRNA. I developed a Cancer-specific RNA-editing Identification using Somatic variation Pipeline (CRISP: github.com/Sasanh/CRISP) a computational framework for accurate identification of A-to-I editing events contributing to the prognosis and stratification of glioblastoma subtypes as well as the editing events that can serve as molecular classifiers for therapeutic approaches. I proposed two models that explains the cis-regulatory role of A-to-I editing events in noncoding regions in modulating the post-transcriptional regulation of target transcripts in glioblastoma.