Although increasingly emphasised in the recent GWAS literature, it is worth noting that the loss of accuracy problem is not utterly new. Indeed, a number of studies in the animal breeding literature have previously reported lower accuracy of genomic selection across genetically distant breeds4,5, consistent with the observation of limited transferability of GWAS findings across diverse human populations6,7. These studies also highlight major factors influencing that loss such as differences between populations in causal variants effect sizes, in alleles frequencies and in linkage disequilibrium (LD) between causal variants and SNPs assayed in GWAS6,8,9. To illustrate the latter point, let us consider a SNP which has an LD r2 with a causal variant of 0.8 in the discovery population and 0.6 in the target population. Such a SNP would therefore explain 25% = (1 − 0.6/0.8) less trait variation and thus be less predictive in the target population as compared with the discovery population, even when causal variants and their effect sizes are shared between ancestries. More generally, previous empirical and simulation studies have shown that accuracy of genetic predictors decays
