Description
Dr. Gunn will discuss findings from a study leveraging data from Million Veterans Program and All of Us Research Program to compare methods for building polygenic scores (PGS) for multi-ancestry populations across multiple traits.
Overview of Presentation
- Polygenic scores (PGS) are a promising tool for identifying people at high genetic risk of disease.
- However, PGS performance declines when scores are applied to target populations different from which they were derived, and most PGS were built with data from primarily European ancestry populations.
- Our study investigates how to best build PGS for diverse, multi-ancestry populations, using GWAS results from the Million Veterans Program (MVP).
- We built polygenic scores (PGS) for ten complex traits using popular single and multi-ancestry Bayesian methods and evaluated these scores in the All of Us Research Program.
- Overall, we conclude that approaches which combine GWAS results from multiple populations produce scores that perform better than single-population approaches.
- Our results contribute to the growing consensus that leveraging GWAS results from multiple-ancestry groups improves PGS performance in populations historically underrepresented in GWAS.
