Empowering gene discovery and accelerating clinical translation for admixed populations

Gene discovery has revolutionized medical genetics, but current methods remain limited in their ability to analyze individuals with complex ancestry profiles. As the U.S. population includes a substantial proportion of individuals with multiple ancestral components, improving analytic methods is essential for maximizing the utility of large-scale research efforts such as the All of Us Research Program. This effort will develop robust, ancestry-aware frameworks for gene discovery and clinical translation, producing scalable publicly available tools and resources that can be integrated into genomic pipelines, facilitating use across research and clinical settings. By improving analytic precision across complex ancestry profiles, this project supports more reproducible genetic findings. Dr. Atkinson proposes to address this issue by developing a suite of innovative statistical methods, software packages, and analytical resources to improve the study of complex traits in individuals with admixed genomes. In this 5-year R01 award, she will: 1a) build a novel statistical method to allow for the integration of admixed individuals into rare variant association studies; 1b) test this new method in a scalable cloud-based software implementation across phenotypes of varying genetic architectures in the All of Us Research Program; and 1c) leverage patterns of linkage disequilibrium in admixed individuals to improve fine-mapping of top loci. She will also develop novel software tools to 2a) realistically simulate ancestrally heterogeneous cohorts; which will be used to 2b) define best practice recommendations for multi-ancestry phasing, imputation, and local ancestry inference; and 2c) assess the impact of common analytic strategies for varied collections on gene discovery outcomes. Finally, she will leverage admixture to 3a) extend ancestry-informed frameworks to quantify gene-gene interactions both locally and distally; and 3b) characterize trends in allelic effect size differences across ancestry components with control over the environment. These efforts fill a gap in existing resources and enhance the accuracy of genetic analyses, supporting gene discovery and clinical translation across genetic backgrounds. All software and simulation tools will be released as open-source packages with detailed documentation, supporting transparency, reproducibility, and adoption. The proposed work aligns with the mission of NIH/NHGRI to advance genomic science and improve health through data-driven innovation.