As observed in Figure 5, more recent GWAS have gathered larger sample sizes and unveiled associations with lower ORs. Although more recent GWAS present decreased replicability rates, an interesting inference can be made by observing effective replicability rates, the ratio between the proportion of positive replications and their statistical power. Effective replicability would be expected to decrease if the lower ORs detected by GWAS were due to lower-frequency (and thus increasingly population-specific) causal variants. In contrast, we observed a constant effective replicability rate of ∼80% that was independent of the OR reported in the European discovery GWAS (red line in Figure 5), indicating that the associations discovered by larger GWAS present similarly high replicability rates regardless of their weaker effect size.
