A0331
Title: Robust multi-ancestry PWAS utilizing Bayesian fine-mapping
Authors: Chengli Zhang - City University of Hong Kong (Hong Kong)
Haoran Xue - City University of Hong Kong (Hong Kong) [presenting]
Chong Wu - The University of Texas MD Anderson Cancer Center (United States)
Abstract: Proteome-wide association studies (PWAS) have emerged as a powerful tool for identifying proteins associated with complex diseases, which can serve as potential drug targets. The conventional PWAS approach employs a two-stage least squares(2SLS) regression, utilizing genetic variants as instrumental variables (IVs). However, the validity of this approach can be compromised by the widespread pleiotropy of genetic variants, which can lead to the identification of false-positive causal proteins for diseases. Furthermore, the varying linkage disequilibrium (LD) patterns and effect sizes of genetic variants across ancestries limit the power of PWAS when analyzing different populations separately. To address these challenges, a robust and powerful PWAS method is proposed that integrates proteomics and disease data from multiple ancestries and employs a Bayesian fine-mapping approach to detect invalid IVs. The effectiveness of the proposed method is demonstrated by applying it to a large-scale biobank dataset, identifying putative causal proteins for complex human diseases.
