When using the trans-ancestry PRS as a predictor of T2D status for eMERGE individuals of African ancestry, the variance explained on the liability scale was 2.8%, the covariates-adjusted AUC was 0.58, and the OR/SD was 1.54 (95% CI: 1.46–1.64; Table 2; Additional File 2: Table S3). As expected, the prediction accuracy of the PRS in the African population was substantially lower compared with the prediction in the European population, reflecting the current Eurocentric bias in genomic studies and the fact that our trans-ancestry PRS was constructed from GWAS of predominantly European and East Asian individuals. Nevertheless, the T2D risk increased with higher PRS values, and the OR comparing individuals in the top 2% of the PRS distribution with the rest of the individuals was 1.98 (95% CI: 1.43–2.74; P value = 4.34E−05), indicating that the trans-ancestry PRS can be useful in identifying individuals with elevated T2D risk in the African population (Table 2; Additional File 2: Table S3).
