Large-scale imputation models for multi-ancestry proteome-wide association analysis

Abstract

Proteome-wide association studies (PWAS) decode the intricate proteomic landscape of biological mechanisms for complex diseases. Traditional PWAS model training relies heavily on individual-level reference proteomes, thereby restricting its capacity to harness the emerging summary-level protein quantitative trait loci (pQTL) data in the public domain. Here we introduced a novel framework to train PWAS models directly from pQTL summary statistics. By leveraging extensive pQTL data from the UK Biobank, deCODE, and ARIC studies, we applied our approach to train large-scale European PWAS models (totaln= 88,838 subjects). Furthermore, we developed PWAS models tailored for Asian and African ancestries by integrating multi-ancestry summary and individual-level data resources (totaln= 914 for Asian and 3,042 for African ancestries). We validated the performance of our PWAS models through a systematic multi-ancestry analysis of over 700 phenotypes across five major genetic data resources. Our results bridge the gap between genomics and proteomics for drug discovery, highlighting novel protein-phenotype links and their transferability across diverse ancestries. The developed PWAS models and data resources are freely available atwww.gcbhub.org.