A Bayesian hierarchical variable selection prior for pathway‐based GWAS using summary statistics

Abstract

While genome‐wide association studies (GWASs) have been widely used to uncover associations between diseases and genetic variants, standard SNP‐level GWASs often lack the power to identify SNPs that individually have a moderate effect size but jointly contribute to the disease. To overcome this problem, pathway‐based GWASs methods have been developed as an alternative strategy that complements SNP‐level approaches. We propose a Bayesian method that uses the generalized fused hierarchical structured variable selection prior to identify pathways associated with the disease using SNP‐level summary statistics. Our prior has the flexibility to take in pathway structural information so that it can model the gene‐level correlation based on prior biological knowledge, an important feature that makes it appealing compared to existing pathway‐based methods. Using simulations, we show that our method outperforms competing methods in various scenarios, particularly when we have pathway structural information that involves complex gene‐gene interactions. We apply our method to the Wellcome Trust Case Control Consortium Crohn's disease GWAS data, demonstrating its practical application to real data.

A Bayesian hierarchically structured prior for gene‐based association testing with multiple traits in genome‐wide association studies

“…The HSVS‐M prior is structured as in which we use , a binary indicator, on for the hypothesis testing. The value of indicates whether we reject the null hypothesis and is considered as a Bayesian analog to a p value and closely related to the Bayes factor (Yang et al, 2020).…”

“…Compared to the conventional approach that tests single‐nucleotide polymorphisms (SNPs) individually, gene‐based tests are often more powerful due to the reduced burden of multiple testing and the information borrowed from SNPs in linkage disequilibrium (LD). However, current gene‐based tests are mostly limited to univariate analysis (Pan et al, 2014, 2015; Wu et al, 2011; Yang et al, 2018, 2020, 2021) that tests the association of a gene with only one trait at a time. Multiple correlated traits are often measured and studied together in the context of complex diseases.…”

“…In this study, we propose the multivariate hierarchically structured variable selection (HSVS‐M) model for testing the association of a gene with multiple traits. The HSVS framework was previously considered for univariate gene‐ and pathway‐based GWASs using summary statistics (Yang et al, 2018, 2020) and genotype data (Yang et al, 2021). We extend the HSVS framework in this study to multivariate gene‐based GWASs that require only summary statistics, which are typically meant for common (minor allele frequency [MAF] 5%) and low‐frequency (1% MAF 5%) variants.…”

“…However, current gene-based tests are mostly limited to univariate analysis (Pan et al, 2014(Pan et al, , 2015Wu et al, 2011;Yang et al, 2018Yang et al, , 2020Yang et al, , 2021 that tests the association of a gene with only one trait at a time. Multiple correlated traits are often measured and studied together in the context of complex diseases.…”

A Bayesian hierarchically structured prior for rare‐variant association testing

“…In this study, we propose the adaptive hierarchically structured variable selection (HSVS-A) prior using a pairwise weighting scheme for set-based rare-variant association tests for both continuous and dichotomous traits. The HSVS framework was previously considered for gene-and pathway-based common variants association tests (Yang et al, 2018(Yang et al, , 2020. We extend and develop in this paper the HSVS framework for testing rare variants.…”