Model Selection in Genome-Wide Association Studies

• Activity Number: 695
• Type: Contributed
• Date/Time: Thursday, August 13, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Kevin Keys* and Gary Chen and Kenneth Lange
• Companies: UCLA and University of Southern California and UCLA
• Keywords: high-dimensional regression ; genome-wide association studies ; nonconvex optimization ; parallel computing ; model selection ; computational statistics

Abstract

Case-control genome-wide association studies (GWASes) examine genetic variation between two groups of patients distinguished by a measurable phenotype. Each patient is genotyped at several places on the genome, and the genetic variation between cases and controls is compared. Since the introduction of GWASes in 2005, researchers have performed GWASes for hundreds of traits on thousands of individuals. GWASes produce massive quantities of data that present computational and model selection challenges to their analysis. We exploit iterative hard thresholding to effectively select genetic markers informative for the trait. Our parallel implementation exploits GPUs and PLINK data compression to fit GWAS data in computer memory. Preliminary tests suggest that our implementation effectively controls type I errors better than LASSO and MCP regression. By leveraging commodity laptop computer hardware, we enable GWAS analysis on desktop machines and discard the need for expensive supercomputing resources.