An Empirical Bayesian Approach for Prioritizing Somatic Mutations

• Activity Number: 532
• Type: Invited
• Date/Time: Wednesday, August 6, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Caucus for Women in Statistics
• Author(s): Keegan Korthauer and Christina Kendziorski*+
• Companies: University of Wisconsin-Madison and University of Wisconsin-Madison
• Keywords: cancer genomics ; TCGA ; empirical Bayes

Abstract

Identifying and prioritizing somatic mutations is an important and challenging area of cancer research that can provide new insights into gene function as well as new targets for drug development. Most methods for prioritizing mutations rely primarily on frequency-based criteria, where a gene is identified as having a causal mutation if it is altered in significantly more samples than expected according to a background model describing random mutations. Once assumed constant across the genome, background models have been improved to accommodate factors such as mutation type, nucleotide context, and various gene-specific covariates, though there is still room for improvement. Even with an accurate background model, frequency-based methods are limited in that they ignore information concerning the likelihood that a given mutation is functional. Here we develop an integrative model-based approach that uses an improved background mutation model and incorporates both frequency and functional impact criteria for inferring driver gene status. Advantages are demonstrated in simulation studies and in an analysis of data from The Cancer Genome Atlas (TCGA) ovarian project.