Co-occurring expression and methylation QTLs allow detection of common causal variants and shared biological mechanisms

Abstract

Inherited genetic variation affects local gene expression and DNA methylation in humans. Most expression quantitative trait loci (cis-eQTLs) occur at the same genomic location as a methylation QTL (cis-meQTL), suggesting a common causal variant and shared mechanism. Using DNA and RNA from peripheral blood of Bangladeshi individuals, here we use co-localization methods to identify eQTL-meQTL pairs likely to share a causal variant. We use partial correlation and mediation analyses to identify >400 of these pairs showing evidence of a causal relationship between expression and methylation (i.e., shared mechanism) with many additional pairs we are underpowered to detect. These co-localized pairs are enriched for SNPs showing opposite associations with expression and methylation, although many SNPs affect multiple CpGs in opposite directions. This work demonstrates the pervasiveness of co-regulated expression and methylation in the human genome. Applying this approach to other types of molecular QTLs can enhance our understanding of regulatory mechanisms.

Primo: integration of multiple GWAS and omics QTL summary statistics for elucidation of molecular mechanisms of trait-associated SNPs and detection of pleiotropy in complex traits

“…Due to limited sample sizes (74) in the pQTL analysis, only 1 out of the 157 examined breast cancer susceptibility loci was associated with cis-protein abundance levels with high confidence, although the cis-gene expression levels and cis-protein abundances for those loci were often highly correlated with an averaged (Pearson) correlation coefficient of r=0.398 and a median of r=0.414. There were 21 out of 157 susceptibility loci uniquely associated with cis-methylation levels but not expression levels in either tumor or normal tissue, echoing a recent work showing both unique and shared causal mechanisms of epigenome variations and transcription [21]. This also shows that the integration of GWAS and multi-omics traits can provide additional insights in understanding the complex and dynamic mechanisms.…”

“…These methods have identified known and novel candidate genes underlying psychiatric disorders [16], diabetes traits [18], obesity-related traits [15,17,19], and others. By applying the integrative methods to multi-omics data, some QTL pairs such as eQTL and methylation (me)QTL pairs have also been identified with evidence of a shared causal mechanism [16,21]. Integrating studies of multiple complex and omics traits could produce a more comprehensive picture of how cellular processes contribute to variation in complex traits.…”

Eliciting priors and relaxing the single causal variant assumption in colocalisation analyses

“…The studies covered a variety of trait pairs, generally integrating a disease GWAS with molecular quantitative trait loci (QTL) data, 20-39 but also comparing pairs of disease GWAS, 40 eQTL and pQTL 41,42 or eQTL and other molecular traits. 43,44 Conditioning was used to allow for multiple causal variants in only one study 40 and 22 out of 25 studies used the software default priors across this diverse range of trait pairs.…”

MOSTWAS: Multi-Omic Strategies for Transcriptome-Wide Association Studies

“…This result accords with Boyle et al's omnigenic model, proposing that regulatory networks are so interconnected that a majority of genetic variants in the genome, local or distal, have indirect effects on transcription of any particular gene [6,7]. Many groups have leveraged this model to identify distal expression quantitative trait loci (eQTLs) by testing the effect of a distal-eSNP on an eGene mediated through a set of genes local to the SNP, concluding that many distal-eQTLs are often eQTLs for many local genes [8][9][10][11][12][13]. It has been shown previously that distal-eQTLs found in regulatory hotspots are generally cell-type specific [8,12,14].…”

“…Many groups have leveraged this model to identify distal expression quantitative trait loci (eQTLs) by testing the effect of a distal-eSNP on an eGene mediated through a set of genes local to the SNP, concluding that many distal-eQTLs are often eQTLs for many local genes [8][9][10][11][12][13]. It has been shown previously that distal-eQTLs found in regulatory hotspots are generally cell-type specific [8,12,14]. Deep learning methods have employed similar logic to link GWAS-identified variants to nearby regulatory mechanisms for functional hypothesis generation [15].…”