Single locus theory of admixture is insufficient for the study of complex traits in admixed populations

Abstract

Admixed populations offer valuable insights into the genetic architecture of complex traits. Investigation of such a topic demands a delicate theory that can handle the complex demographic history of admixed populations. With additional assumptions, we show that Pritchard-Stephens-Donnelly (PSD) can predict empirical GWAS findings. We name this the extended PSD (ePSD) model and evaluate the predictions using real data. When applied to real data of admixed genomes, the prediction of the ePSD model is very successful in explaining single-locus behaviors while falling short in predicting two-locus phenomena involving linkage disequilibrium. Our results show that a mosaic of independent single-continental segments is an insufficient approximation of contemporary admixed populations. A more advanced theory that better models linkage disequilibrium of admixed populations will be crucial to better understanding the genetic architecture of complex traits.