Kernel-Machine Methods for Gene Mapping of Complex Traits

• Activity Number: 215
• Type: Invited
• Date/Time: Monday, August 2, 2010 : 2:00 PM to 3:50 PM
• Sponsor: ENAR
• Author(s): Michael Epstein*+
• Companies: Emory University
• Keywords: association ; gene ; kernel

Abstract

Genetic association studies of complex traits often genotype a set of SNPs that summarize genetic variation across the genome and then tests for association between each SNP and trait using a univariate test. Such tests can lose power if the tested SNP is in incomplete linkage disequilibrium with the trait locus. Alternatively, one can test multiple SNPs across a specific region (e.g. gene) using multivariate methods but such tests have large degrees of freedom that can also compromise power. In this talk, I propose the use of kernel-machine methods for association mapping that uses information from multiple SNPs simultaneously in analysis but produces a test statistic with reduced degrees of freedom compared to multivariate approaches. Using both simulated and real data, we show kernel-machine methods provide a novel and powerful strategy for gene mapping.