A Beta-Binomial Model to Compare Somatic Mutation Rates Between Groups of Cancer Patients

• Activity Number: 252
• Type: Contributed
• Date/Time: Monday, August 1, 2016 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Genomics and Genetics
• Author(s): Hong Wang* and Yong Chen and Jinze Liu and Heidi Weiss and Susanne M. Arnold and Aronld Stromberg and Chi Wang
• Companies: University of Kentucky and University of Pennsylvania Perelman School of Medicine and University of Kentucky and University of Kentucky and University of Kentucky and University of Kentucky and University of Kentucky
• Keywords: Beta-Binomial ; Somatic Mutation ; Whole-exome sequencing ; Likelihood ratio test ; Genomic

Abstract

Whole-exome sequencing (WES) provides a powerful approach to profile somatic gene mutations in cancer genome. A fundamental question in the analysis of WES data is how to compare somatic mutation patterns between groups of patients with varying characteristics such as geographic region, tumor stage, or response to therapy. Current methods, such as the Fisher's exact test, do not recognize a) different types of mutations, b) do not account for the background mutation rate, and c) do not adjust for demographic and clinical covariates. As those methods oversimplify the comparison problem, they may lead to less than optimal results. In this talk, we present a beta-binomial model-based approach to compare somatic mutations between patient groups while fully accounting for the complexities that are overlooked by other methods. Specifically, our model accounts for variations in mutation rate, normalizes data based on background mutation rate, and adjusts for baseline covariates. We propose an empirical Bayes shrinkage approach to estimate the dispersion parameter in the beta-binomial model and a likelihood ratio test to identify differentially mutated genes.