are lacking to address most questions about putative differences in polygenic scores across worldwide populations. The straightforward, albeit expensive and time-consuming, solution to improving polygenic score performance across diverse populations is to create well-powered GWAS data resources for many different worldwide populations. However, the collection of very many samples from ancestrally disparate populations could be a poor use of resources, if all samples are underpowered. Perhaps the best approach is strategic collection of large samples from multiple populations, some of which may be ancestrally homogeneous, while others of which may be from specific admixed populations. This approach would afford cross-ancestry comparisons of more ancestrally homogeneous populations (which are easier to analyze) as well as much-needed method development for admixed populations (which are currently considered to be more difficult to analyze).
