GCPBayes: An R package for studying Cross-Phenotype Genetic Associations with Group-level Bayesian Meta-Analysis

Abstract

Several R packages have been developed to study cross-phenotypes associations (or pleiotropy) at the SNP-level, based on summary statistics data from genome-wide association studies (GWAS). However, none of them allow for consideration of the underlying group structure of the data. We developed an R package, entitled GCPBayes (Group level Bayesian Meta-Analysis for Studying Cross-Phenotype Genetic Associations), introduced by Baghfalaki et al. (2021), that implements continuous and Dirac spike priors for group selection, and also a Bayesian sparse group selection approach with hierarchical spike and slab priors, to select important variables at the group level and within the groups. The methods use summary statistics data from association studies or individual level data as inputs, and perform Bayesian meta-analysis approaches across multiple phenotypes to detect pleiotropy at both group-level (e.g., at the gene or pathway level) and within group (e.g., at the SNP level).