Regarding scientific and public perception of polygenic scores, it is important to address apparent differences in polygenic score distributions across populations. Our findings suggest that it is currently not possible to know precisely the distribution of polygenic scores for non-European ancestry populations, for any complex genetic phenotype, for two major reasons: (1) score distributions depend critically on methodological choices in score constructions, and (2) data resources for most populations are currently inadequate. Further, as we have shown, the ordering of population distributions of polygenic scores varies under accepted methods of constructing these scores (i.e., using different p-value thresholds for variant inclusion in scores and using alternative discovery GWAS). Explanations for these differences are currently incomplete. Until vastly superior data resources are available—including large-scale GWAS in multiple globally representative populations—scientists are unlikely to reach consensus regarding the existence, nature, and exact causes of polygenic score differences among populations.
