To create a multi-ancestry PRS without incorporating local ancestry, Márquez-Luna et al.10 use a mixture of PRSs, taking advantage of existing well-powered GWAS studies and supplementing with additional information that can be gained from a smaller study in the population of interest. We investigate this approach in the context of varying admixture proportions and find that it achieved high accuracy across all admixed individuals, was not biased by ancestry, and significantly improved performance over a European-only PRS with 10-fold fewer African ancestry cases. Thus, a combination of multiple single-population PRSs may be the best currently available approach for admixed individuals, and this approach will likely continue to improve as the individual PRSs are further developed.
