A1351
Title: Scalable multi-trait fine-mapping for metabolite GWAS
Authors: Weiqiong Huang - Department of Statistics, University of Pittsburgh, PA, US (United States) [presenting]
Christopher McKennan - University of Pittsburgh (United States)
Joshua Cape - University of Wisconsin, Madison (United States)
Emily Hector - North Carolina State University (United States)
Abstract: Genome-wide association studies (GWAS) have enabled the discovery of thousands of genetic loci associated with complex traits, yet identifying the causal variants and understanding their biological mechanisms requires statistical fine-mapping. Existing fine-mapping approaches face two major limitations: They are often computationally prohibitive when scaling to hundreds of traits, and they typically assume independence of genetic effects, which can lead to biased inference in the presence of pleiotropy and linkage disequilibrium. A Bayesian factor model is introduced for scalable and interpretable multi-trait fine-mapping from GWAS summary statistics. The model decomposes genetic effects into shared components mediated by latent biological processes and trait-specific effects, allowing for the capture of both pleiotropic and trait-unique signals. Inference procedure leverages sequential approximation and efficient model space exploration, achieving orders-of-magnitude speedups over conventional methods while retaining statistical rigor. Crucially, the framework naturally accommodates biologically informed priors such as metabolic pathway structures in metabolomics to interpretable inference grounded in domain knowledge. Applying the method to metabolite GWAS of over 700 traits, findings from studies with 20 larger sample sizes are replicated, and novel causal variants that are inaccessible to current fine-mapping pipelines are uncovered.
