Genetic Ancestry Inference for Pharmacogenomics

Abstract

Genetic ancestry inference can be used to stratify patient cohorts and to model pharmacogenomic variation within and between populations. We provide a detailed guide to genetic ancestry inference using genome-wide genetic variant datasets, with an emphasis on two widely used techniques: principal components analysis (PCA) and ADMIXTURE analysis. PCA can be used for patient stratification and categorical ancestry inference, whereas ADMIXTURE is used to characterize genetic ancestry as a continuous variable. Visualization methods are critical for the interpretation of genetic ancestry inference methods, and we provide instructions for how the results of PCA and ADMIXTURE can be effectively visualized.

Keywords: Admixture; Genetic ancestry inference; Genetic variants; Health disparities; Pharmacogenomics; Population-specific drug efficacy.

Fig. 1

Principal components analysis (PCA) of four human populations: Colombia (CLM – green), Peru (PEL – red), Spain (IBS – orange), Yoruba (YRI – blue). The first two principal components (PCs) are shown

Fig. 2

ADMIXTURE plot of four human populations: Colombia (CLM), Spain (IBS), Peru (PEL), and Yoruba (YRI). Each column is an individual, and for each individual the ancestry fraction for each of three continental population groups is shown: African (blue), European (orange), and Native American (red)