Extended Data Fig. 8: Genetic risk prediction accuracy in EAS from EAS or EUR training data.
From: Comparative genetic architectures of schizophrenia in East Asian and European populations
As in Fig. 4, PRS shows case/control variance explained with EUR and EAS samples using a leave-one-out meta-analysis approach for the EAS samples. Error bars indicate the 95% confidence intervals. a,b, Liability-scale variance explained when LD panel for clumping is from EUR 1000 Genomes Phase 3 samples and best-guess genotypes are from each EAS cohort. c,d, Nagelkerke’s R2 for EAS prediction accuracy when LD panel for clumping is from EUR and EAS 1000 Genomes Phase 3 samples. e,f, Nagelkerke’s R2 for EAS prediction accuracy when LD panel for clumping is from EUR 1000 Genomes Phase 3 samples and best-guess genotypes are from each EAS cohort. a-f, EAS stage 1 (n = 13,305 cases; 16,244 controls) and EUR samples (n = 33,640 cases; 43,456 controls).
