Skip to main content
. 2019 Mar 26;35(20):4038–4044. doi: 10.1093/bioinformatics/btz176

Fig. 4.

Fig. 4.

AUC values and their standard errors for PRS of schizophrenia for genotyped SNPs in validation dataset. PRSs were trained based on the PGC data excluding the MGS study; the MGS study was used as the validation data for calculating AUC and its variance. The x-coordinate is the number of SNPs (with smallest P-values in the training dataset) for PRS. AUC0 and SE0 were calculated assuming individual level data. AUC1 and SE1 were calculated using SummaryAUC based on GWAS summary data in MGS. r2=0.01: SNPs were LD-clumped using r2=0.01 in PLINK. r2=0.2: SNPs were LD-clumped using r2=0.2 in PLINK