Of course, the finding of shared variants underlying GWAS results holds only for associations that have been published so far. Ongoing efforts to join cohorts into large consortia [34] ensure steady progress in the field and guarantee the discovery of new genetic associations to complex disease [6], [35]. It is tempting to make inferences about what may be the results of future, much larger, association studies; particularly about the frequency and degree of trans-ethnic sharing of as yet undiscovered variants. We approximated this question by considering the allele frequencies and effect sizes of associated SNPs along with their patterns of replicability across time. Specifically, it is clear that if the GWAS with larger sample sizes that have been published recently for peoples of European ancestry had discovered variants with lower frequencies (variants that should be increasingly population-specific), their results should be less likely to replicate across populations. If this observation were made, it would predict decreased replicabilities in future, even larger GWAS with increased power to discover lower-frequency risk variants.
