Covariate-Modulated Local False Discovery Rate for Incorporating SNP Annotations in Genome-Wide Association Studies

• Activity Number: 62
• Type: Topic Contributed
• Date/Time: Sunday, July 29, 2012 : 4:00 PM to 5:50 PM
• Sponsor: Biometrics Section
• Author(s): Rong Zablocki*+ and Richard A. Levine and Wesley Thompson
• Companies: University of California at San Diego and San Diego State University/JCGS and University of California at San Diego
• Keywords: GWAS ; CM-LFDR ; SNP ; mixture model ; annotation ; heritability

Abstract

Traditional genome-wide association studies (GWAS) evidenced missing heritability arising from a large number of small polygenic effects in heritable complex phenotypes; the Bonferoni correction was over conservative. Schork et al. (2012) proposed a novel approach incorporating the concept of covariate-modulated local false discovery rate (CM-LFDR) into the mixture model of a weighted linear combination of a null density of single nucleotide polymorphism (SNP) without trait-association, and a non-null density with trait-association; information regarding the allocation and linkage disequilibrium (LD) of SNPs were handled as covariates. Hence, a common genetic architecture across diverse complex traits emerged; the regulatory and exonic variation within protein coding genes explained more phenotypic variance per SNP, and replicated at higher rates across independent samples. The goal of the current study is to improve the model performance genetically and/or computationally by adding other potential covariates (e.g., annotations) and seek alternative computational strategies (e.g., variational approximation).