Epigenome-wide association study and integrative analysis with the transcriptome based on GWAS summary statistics

TL;DR

This paper introduces a Mendelian randomization-based framework for epigenome-wide association studies that integrates methylation and transcriptome data with GWAS summary statistics, uncovering novel disease-related genes.

Contribution

It presents a new general framework for integrative analysis of GWAS, methylation, and transcriptome data, revealing novel candidate genes and disease mechanisms.

Findings

Identified loci with differential methylation linked to genetic variations.

Discovered novel candidate genes not previously associated with GWAS.

Strong evidence of differential expression among top methylation-linked genes.

Abstract

The past decade has seen a rapid growth in omics technologies. Genome-wide association studies (GWAS) have uncovered susceptibility variants for a variety of complex traits. However, the functional significance of most discovered variants are still not fully understood. On the other hand, there is increasing interest in exploring the role of epigenetic variations such as DNA methylation in disease pathogenesis. In this work, we present a general framework for epigenome-wide association study and integrative analysis with the transcriptome based on GWAS summary statistics and data from methylation and expression quantitative trait loci (QTL) studies. The framework is based on Mendelian randomization, which is much less vulnerable to confounding and reverse causation compared to conventional studies. The framework was applied to five complex diseases. We first identified loci that are differentially methylated due to genetic variations, and then developed several approaches for joint testing with the GWAS-imputed transcriptome. We discovered a number of novel candidate genes that are not implicated in the original GWAS studies. We also observed strong evidence (lowest p = 2.01e-184) for differential expression among the top genes mapped to methylation loci. The framework proposed here opens a new way of analyzing GWAS summary data and will be useful for gaining deeper insight into disease mechanisms.