Bayesian Unification of Genetics and Epigenetics to Understand Regulation, Diseases and Development

Talk
Avinash Das Sahu
Time: 
11.04.2014 11:00 to 12:30
Location: 

CBCB 3118

In my thesis, I propose to develop novel Bayesian methods based computational approaches to deepen the understanding of human biology.
At first we address the problem of genotype-phenotype association. While numerous Genome-Wide Association Studies (GWAS) pertaining to genetic diseases and expression Quantitative Trait Loci (eQTL) studies have revealed hundreds of genomic variants associated with phenotype or gene expression, the gap between statistical associations and mechanistic translation of genotype into phenotype remains daunting. The reasons for this are primarily linkage disequilibrium in the genome, which confounds causality with mere association, and poor understanding of functional implications of non-coding variants in the genome. I propose to overcome these challenges by prudently leveraging the recently available epigenetic data, which enables us to functionally annotate the non-coding genome, and by developing Bayesian, mechanism-based, model of genotype-phenotype associations. The model is based on the premise that non-coding variants in distal enhancer regions jointly affect expression of specific gene, which, in turn, perturbs downstream pathways and processes, manifesting as a specific disorder.
Second set of problem addressed in the thesis concern regulatory dynamics of spatiotemporal development. Even with wide availability of spatiotemporal regulatory data, lack of statistical framework for its analysis hinders identification of mechanism underlying tissue differentiation. I propose to develop a novel statistical framework that can efficiently identify specific spatiotemporal events that help us to understand regulatory dynamics in development and tissue diversity. We will apply our models to publicly available data, as well as data generated by our collaborators in the MagNET consortium, to decipher regulatory networks, disease causing variants and the underlying mechanism. Overall, the proposed research will result in several novel biologically relevant and theoretically sound computational approaches and corresponding software tools for understanding diseases and development in humans.
Examining Committee:
Committee Chair: - Dr. Sridhar Hannenhalli
Dept's Representative - Dr. Hal Daume' III
Committee Member(s): - Dr. Laura Elnitski
- Dr. Hector Corrada Bravo
- Dr. Eytan Ruppin