Multivariate Score Tests in GWAS of Binary Traits in Related Populations with Risk Covariates: The GLOMS System for Gene-Wise, GxE, and GxG Analyses

• Activity Number: 623
• Type: Contributed
• Date/Time: Thursday, August 2, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Stephen Stanhope*+ and Mark Abney
• Companies: The University of Chicago and The University of Chicago
• Keywords: Genome-wide association studies ; Mixed models ; Score test ; High performance computing

Abstract

Currently, substantial attention is being paid to a variety of topics related to GWAS in related populations, the use of covariate information in GWAS, and methods for performing related analyses that combine information from genetic markers with each other or external environmental data. Often such analyses are considered in isolation from one another. For example, a system for performing marker-wise GWAS might be regarded as distinct from one for aggregating information and performing a gene-wise analysis, or a system for performing a gene-wise analysis might be regarded as distinct from one for analyzing GxE interactions.

In this talk, we discuss our efforts to unite such strands of research, and develop effective and tractable methods for performing gene-wise, GxE, and GxG studies of genome-wide genotype data. We describe GLOMS (Genome-wide LOgistic mixed model / Multivariate Score test), a parallelized, computationally efficient and powerful system for such analyses. We demonstrate its performance in simulation studies, and show how GLOMS can be applied to studies of hypertension in the Hutterites, a population related though a 13-generation pedigree.