Analysis of Polygenic Score Usage and Performance across Diverse Human Populations

Studies of the relationship between genetic and phenotypic variation have historically been carried out on people of European ancestry. Efforts are underway to address this limitation, but until they succeed, the legacy of a Euro-centric bias in medical genetic studies will continue to hinder research, including the use of polygenic scores, which are individual-level metrics of genetic risk. Ongoing debate surrounds the generalizability of polygenic scores based on genome-wide association studies (GWAS) conducted in European ancestry samples, to non-European ancestry samples. We analyzed the first decade of polygenic scoring studies (2008-2017, inclusive), and found that 67% of studies included exclusively European ancestry participants and another 19% included only East Asian ancestry participants. Only 3.8% of studies were carried out on samples of African, Hispanic, or Indigenous peoples. We find that effect sizes for European ancestry-derived polygenic scores are only 36% as large in African ancestry samples, as in European ancestry samples (t=-10.056, df=22, p=5.5x10-10). Poorer performance was also observed in other non-European ancestry samples. Analysis of polygenic scores in the 1000Genomes samples revealed many strong correlations with global principal components, and relationships between height polygenic scores and height phenotypes that were highly variable depending on methodological choices in polygenic score construction. As polygenic score use increases in research, precision medicine, and direct-to-consumer testing, improved handling of linkage disequilibrium and variant frequencies (both of which currently reduce transferability of scores) across populations will improve polygenic score performance. These findings bolster the rationale for large-scale GWAS in diverse human populations.