Imputation-based Genetic Association Analysis of Complex Traits in Admixed Populations Public Deposited

Downloadable Content

Download PDF
Last Modified
  • March 21, 2019
  • Duan, Qing
    • Affiliation: School of Medicine, Curriculum in Bioinformatics and Computational Biology
  • Genetic association studies in admixed populations have drawn increasing attention from the genetic community, as performing association analysis in diverse populations allows us to gain deeper understanding of the genetic architecture of human diseases and traits. However, population stratification due to admixture poses special challenges. To address the challenges, I conducted the following studies from the perspectives of enhancing genotype imputation quality and providing proper treatment of local ancestry in the association analysis. First, I provided a new resource of marker imputability information with commonly used reference panels to guide the choice of reference and genotyping platforms. To be specific, I systematically evaluated marker imputation quality using sequencing-based reference panels from the 1000 Genomes Project and released the information through a user-friendly and publicly available data portal. This is the first resource providing variant imputability information specific to each continental group and to each genotyping platform. Second, I established a paradigm for better imputation in African Americans using study-specific sequencing based reference panels. I built an internal reference panel consisting of variants derived from the NHLBI Exome Sequencing Project for African American subjects, which significantly increased effective sample size comparing with that from the 1000 Genomes Project. No loss of imputation quality was observed using a panel built from phenotypic extremes. In addition, I recommended using haplotypes from Exome Sequencing Project alone or concatenation of the two panels over quality score-based post-imputation selection or IMPUTE2’s two-panel combination. Finally, I proposed a robust and powerful two-step testing procedure for association analysis in admixed populations. Through extensive numeric simulations, I demonstrated that our testing procedure robustly captures and pinpoints associations due to allele effect, ancestry effect or the existence of effect heterogeneity between the two ancestral populations. In particular, our testing procedure is more powerful in identifying the presence of effect heterogeneity than traditional cross-product interaction model. I further illustrated its usefulness by applying the two-step testing procedure to test for the association between genetic variants and hemoglobin trait in African American participates from CARe. Taken together, the above studies guide genotype imputation practice and substantially improve the power of imputation-based genetic association studies in admixed populations, leading to more accurate discovery of disease-associated variants and ultimately better therapeutic strategies in admixed populations.
Date of publication
Resource type
Rights statement
  • In Copyright
  • Li, Yun
  • Furey, Terrence
  • Lange, Ethan
  • Mohlke, Karen
  • North, Kari
  • Doctor of Philosophy
Degree granting institution
  • University of North Carolina at Chapel Hill Graduate School
Graduation year
  • 2016

This work has no parents.