A New Statistical Method for Genome-Scale Mutual Exclusivity Analysis of Tumor Mutations

• Activity Number: 679 - Variable Selection and Prediction Models for Genomic Data
• Type: Contributed
• Date/Time: Thursday, August 2, 2018 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Genomics and Genetics
• Author(s): Chi Wang* and Sisheng Liu and Jinpeng Liu and Thilakam Murali and Tingting Zhai and Li Chen and Arnold Stromberg and Hunter Moseley
• Companies: University of Kentucky and Fred Hutchinson Cancer Research Center and Markey Cancer Center, University of Kentucky and University of Kentucky and University of Kentucky and University of Kentucky and University of Kentucky and University of Kentucky
• Keywords: Tumor mutations; Mutual exclusivity analysis; sequencing data analysis

Abstract

Mutual exclusivity analysis of tumor mutation sequencing data has been shown as an effective approach to identify cancer driver genes. Although many statistical and bioinformatics methods have been proposed to identify mutual exclusive mutational patterns, there are three major limitations. First, the heterogeneity in background mutation rates across patients have not been taken into account. Second, a highly mutated gene may dominant a mutation pattern and lead to false positive results. Third, a genome-scale search for large gene sets representing longer pathways is infeasible due to the computational intensity of statistical tests. In this paper, we propose a new statistical test for mutual exclusivity that 1) incorporates the background mutation rate for each gene in each patient; 2) balances the contribution of each gene in a gene set to the mutual exclusive pattern; 3) is very fast to calculate. Combined with an efficient Markov Chain Monte Carlo algorithm, our method enables genome-wide screening of mutual exclusive mutational patterns. The method has been applied to somatic mutation data from The Cancer Genome Atlas.