Regarding expectations about polygenic score performance in non-European ancestry samples we note that polygenic scores for many complex genetic phenotypes are strongly correlated with global PCs, which highlights the critical importance of appropriate statistical methods for the analysis of genetic data from admixed populations, and caution in using imputed data from samples for which reference sequence datasets are lacking. Testing future polygenic scoring results for robustness to the inclusion of variable combinations of PCs will reduce the chances of spurious results (that are actually attributable to ancestry). In sum, the preponderance of genetic studies based on European ancestry samples has led to a situation in which polygenic scores are approximately one-third as informative for African ancestry individuals, as they are for European ancestry individuals. This is presumably true for commercially available tests as well; given these findings, consumers should be aware of the differential performance of tests across individuals.
