Fig. 2: Accuracy of PP-GWAS against REGENIE (centralized) on real-world datasets.
From: PP-GWAS: Privacy Preserving Multi-Site Genome-wide Association Studies
A Scatter (correlation) plot for the Bladder Cancer Risk dataset, showing the SNP-wise agreement between \(-{\log }_{10}(p)\) from PP-GWAS and REGENIE. Pearson correlation (two-sided) between \(-{\log }_{10}(p)\) across SNPs (M = 467, 172): r = 1.000000 (95% CI [0.999999, 1.000000]); t(467, 170) = 677, 871.82, P < 10−6; R2 = 0.999999. B Scatter (correlation) plot for the Age-Related Macular Degeneration (AMD) dataset, depicting the analogous correlation. Pearson correlation (two-sided) between \(-{\log }_{10}(p)\) across SNPs (M = 508, 740): r = 1.000000 (95% CI [0.999999, 1.000000]); t(508, 738) = 577, 736.14, P < 10−6; R2 = 0.999998. No multiple-testing correction was applied to these correlation tests.
