To summarize the performance of the trans-ancestry PRS across cohorts, we derived overall OR, and prevalence-adjusted PPV and NPV estimates at various percentage cutoffs for the African population by meta-analyzing the eMERGE African dataset with the four UAB Black cohorts, and overall performance metrics for the East Asian population by meta-analyzing the three TWB batches (Additional File 2: Table S6). Along with the eMERGE European dataset, at least 4500 T2D cases were available for the three populations, enabling an accurate assessment of the tail discrimination of the trans-ancestry PRS. We note that while there was a substantial difference in case prevalence across eMERGE African and UAB Black cohorts, the performance metrics of the trans-ancestry PRS (e.g., AUC, PPV, and NPV) appeared to be robust to this variation. Figure 3 shows that the PRS was highly significantly associated with T2D status across different percentage cutoffs and ancestral groups, with comparable predictive performance in the European and East Asian populations and lower prediction accuracy in the African population. Individuals in the top 2% of the PRS distribution had significantly increased T2D risk,
