Universal genome-wide association studies: Powerful joint ancestry and association testing

• Published in: HGG advances Vol. 4; no. 4; p. 100235
• Main Authors: Shriner, Daniel, Bentley, Amy R., Gouveia, Mateus H., Heuston, Elisabeth F., Doumatey, Ayo P., Chen, Guanjie, Zhou, Jie, Adeyemo, Adebowale, Rotimi, Charles N.
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 12-10-2023; Elsevier
• Subjects: Black or African American - genetics; Gene Frequency; Genome-Wide Association Study; Humans; Linkage Disequilibrium; White
• ISSN: 2666-2477; 2666-2477
• DOI: 10.1016/j.xhgg.2023.100235

The vast majority of human populations and individuals have mixed ancestry. Consequently, adjustment for locus-specific ancestry is essential for genetic association studies. To empower association studies for all populations, it is necessary to integrate effects of locus-specific ancestry and genotype. We developed a joint test of ancestry and association that can be performed with summary statistics, is independent of study design, can take advantage of locus-specific ancestry effects to boost power in association testing, and can utilize association effects to fine map admixture peaks. We illustrate the test using the association between serum triglycerides and LPL. By combining data from African Americans, European Americans, and West Africans, we identify three conditionally independent variants with varying amounts of ancestrally differentiated allele frequencies. Using out-of-sample data, we demonstrate improved prediction achievable by accounting for multiple causal variants and locus-specific ancestry effects at a single locus.
We describe a joint test of ancestry and association calculated from summary statistics from admixture mapping and association studies. Using data for serum triglycerides and LPL from African Americans, European Americans, and West Africans, we identify three conditionally independent variants with varying amounts of ancestrally differentiated allele frequencies.