Center for Medical Ethics and Health Policy Staff Publications
Publication Date
6-2-2023
Journal
Human Genomics
DOI
10.1186/s40246-023-00487-3
PMID
37268996
PMCID
PMC10239111
PubMedCentral® Posted Date
6-2-2023
PubMedCentral® Full Text Version
Post-print
Published Open-Access
yes
Keywords
Humans, Ethnicity, Genome-Wide Association Study, Polymorphism, Single Nucleotide, Veterans, Racial Groups, Ancestry, Race, Ethnicity, Population stratification, Height, Million Veteran Program
Abstract
Background: The Million Veteran Program (MVP) participants represent 100 years of US history, including significant social and demographic changes over time. Our study assessed two aspects of the MVP: (i) longitudinal changes in population diversity and (ii) how these changes can be accounted for in genome-wide association studies (GWAS). To investigate these aspects, we divided MVP participants into five birth cohorts (N-range = 123,888 [born from 1943 to 1947] to 136,699 [born from 1948 to 1953]).
Results: Ancestry groups were defined by (i) HARE (harmonized ancestry and race/ethnicity) and (ii) a random-forest clustering approach using the 1000 Genomes Project and the Human Genome Diversity Project (1kGP + HGDP) reference panels (77 world populations representing six continental groups). In these groups, we performed GWASs of height, a trait potentially affected by population stratification. Birth cohorts demonstrate important trends in ancestry diversity over time. More recent HARE-assigned Europeans, Africans, and Hispanics had lower European ancestry proportions than older birth cohorts (0.010 < Cohen's d < 0.259, p < 7.80 × 10-4). Conversely, HARE-assigned East Asians showed an increase in European ancestry proportion over time. In GWAS of height using HARE assignments, genomic inflation due to population stratification was prevalent across all birth cohorts (linkage disequilibrium score regression intercept = 1.08 ± 0.042). The 1kGP + HGDP-based ancestry assignment significantly reduced the population stratification (mean intercept reduction = 0.045 ± 0.007, p < 0.05) confounding in the GWAS statistics.
Conclusions: This study provides a characterization of ancestry diversity of the MVP cohort over time and compares two strategies to infer genetically defined ancestry groups by assessing differences in controlling population stratification in genome-wide association studies.
Included in
Bioethics and Medical Ethics Commons, Health Policy Commons, Medical Sciences Commons, Race and Ethnicity Commons