Fig 1.
Theoretical predictions of power per causal SNP (upper panel) and out-of-sample R2 of the PGS (lower panel), for total sample size (x-axis) and cross-study genetic correlation (y-axis).
Factor levels: 50 studies, 100k independent SNPs, and arising from a subset of 1k independent SNPs.
Fig 2.
Theoretical predictions of power per causal SNP (upper panel) and out-of-sample R2 of the PGS (lower panel), for a trait that across studies has SNP heritability (x-axis) and cross-study genetic correlation (y-axis).
Factor levels: 50 studies, sample size 5,000 individuals per study, 100k independent SNPs, and heritability arising from a subset of 1k independent SNPs.
Fig 3.
Theoretical predictions of power per causal SNP (upper panel) and out-of-sample R2 of the PGS (lower panel), for a trait with GWAS results from the number of studies (x-axis) with cross-study genetic correlation (y-axis).
Factor levels: total sample size 250,000 individuals, 100k independent SNPs, and arising from a subset of 1k independent SNPs.
Fig 4.
Theoretical predictions of out-of-sample R2 of the PGS, for the SNP heritability in the hold-out sample (x-axis) and the SNP heritability in the discovery samples (y-axis).
Factor levels: 50 studies, sample size 5,000 individuals per study, cross-study genetic correlation 0.8, 100k independent SNPs, and heritability arising from a subset of 1k independent SNPs.
Fig 5.
Theoretical predictions of power per causal SNP, for total sample size (x-axis) and CGR between two sets of studies (y-axis).
Factor levels: 2 sets of 50 studies, CGR equal to 1 within both sets, 100k independent SNPs, and arising from a subset of 1k independent SNPs.
Table 1.
GREML estimates of SNP heritability and genetic correlation across studies and sexes.
Table 2.
Predicted and observed number of genome-wide-significant hits and PGS R2, for large-scale GWAS efforts to date for height, BMI, EduYears, and self-rated health, assuming 250k effective SNPs (i.e., independent haplotype blocks) of which 20k trait-affecting, using averaged GREML estimates from Table 1 for setting SNP heritability and CGR.
Notes on the sources for the large-scale GWAS efforts are listed in S3 Table.