Integrative modeling of multiple genomic data from different types of genetic association studies

Abstract

Genome-wide association studies (GWASs) and expression-/methylation-quantitative trait loci (eQTL/mQTL) studies constitute popular approaches for investigating the association of single nucleotide polymorphisms (SNPs) with disease and expression/methylation, respectively. Here, we propose to integrate QTL studies to more powerfully test the SNP effect on disease in GWASs when they are conducted among different subjects. We propose a model for the joint effect of SNPs, methylation, and gene expression on disease risk and obtain the marginal model for SNPs by integrating out methylation and expression. We characterize all possible causal relations among SNPs, methylation, and expression and study the corresponding null hypotheses of no SNP effect in terms of the regression coefficients in the joint model. We develop a score test for variance components of regression coefficients to evaluate the genetic effect. We further propose an omnibus test to accommodate different models. We illustrate the utility of the proposed method in an asthma GWAS study, a brain tumor study, and numerical simulations.

Keywords: Data integration; Epigenetics; Mediation analysis; Variance component test.