bGWAS_R

bGWAS_R implements Bayesian genome-wide association analyses in R that integrate priors derived from related-trait GWAS via multivariable Mendelian randomization to improve detection of direct genetic effects and disentangle direct versus mediated biological pathways.

Key Features:

• Bayesian Framework: Employs a Bayesian statistical framework to derive informative priors for genetic associations by incorporating prior knowledge about related risk factors and their causal effects on the trait.
• Leveraging Related Studies: Utilizes published GWAS data from related traits to inform prior distributions through multivariable Mendelian randomization estimating causal effect sizes.
• Comprehensive Output: Produces Bayes Factors, posterior distributions/effects, and estimates of direct genetic effects.
• Dissection of Biological Mechanisms: Separates direct and indirect (mediated) effects of genetic variants to clarify biological pathways influencing traits.

Scientific Applications:

• Increasing Statistical Power: Incorporates prior information to enhance power for detecting genetic associations, particularly when sample sizes are limited.
• Understanding Genetic Architecture: Aids in distinguishing direct versus mediated variant effects to inform models of trait etiology.
• Cross-Study Integration: Leverages existing GWAS summary data from related studies to inform priors and enable cross-study analytical integration.

Methodology:

Prior derivation via multivariable Mendelian randomization to obtain causal effect estimates from related traits; Bayesian inference combining these informative priors with observed GWAS data to compute Bayes Factors and posterior distributions; decomposition of effects to separate direct from indirect (mediated) genetic effects.