Project Summary/Abstract Polygenic risk scores (PRS), that aggregate risk across common variants in the genome, have emerged as a powerful tool towards implementing genomic medicine. Unfortunately, the vast majority of genomic data from which current PRS are estimated is coming from European ancestry individuals thus prohibiting the implementation of PRS for non-European individuals. To address this gap, multiple large-scale genomic studies are currently performed in non-European individuals. Of particular interest are individuals with recent ancestry from multiple continental sources such as African Americans and/or Hispanic Americans whose genomes are a mosaic of segments of various ancestries. Such diversity in genetic ancestry raises unique challenges in the equitable PRS development as the accuracy and bias of existing PRS varies across genomic ancestries. Unlike existing paradigm that largely views genetic ancestry as a confounder in PRS studies, we aim to fully integrate population genetics of the admixture process to yield admixture-PRS that provide equitable accuracies for all individuals irrespective of genetic ancestries. We will integrate data of over 230,000 admixed individuals across five diverse medical systems including UCLA, Mt Sinai, Colorado to develop, calibrate and benchmark PRS for admixed individuals.

PROJECT NARRATIVE Although genetic studies of complex human traits have been tremendously successful in identifying thousands of genomic risk regions that harbor variants increasing risk, Unfortunately, the vast majority include individuals of European ancestry making the transferability to individuals with mixed ancestries (e.g., African Americans) challenging. In this proposal we develop and apply new statistical and experimental methods to build admixture-aware polygenic risk scores for complex traits. We leverage a unique data set of >230,000 admixed individuals whose genotypes are linked to electronic health records to develop and calibrate PRS in admixed populations.