Fig. 7: Computational cost of single-ancestry methods.

a Total CPU hours of single-ancestry methods for each trait evaluated in UK Biobank European testing datasets (N = 188,745). X-axis: single-ancestry methods. Y-axis: Total CPU hours (in log scale). b Maximum memory usage of single-ancestry methods of each trait evaluated in UK Biobank European testing datasets (N = 188,745). X-axis: single-ancestry methods. Y-axis: Maximum memory usage (MB). Statistics of computational time and memory for each method are recorded by completing all corresponding parameter settings, except for PRS-CS and SBayesR, which complete one parameter calculation before proceeding to another; thus, the computational cost of PRS-CS and SBayesR was counted by completing only one parameter setting. In each box plot, the central horizontal line represents the median value across traits. The box indicates the interquartile range (IQR, 25th–75th percentiles), and whiskers extend to the most extreme values within 1.5×IQR from the box. Points beyond the whiskers represent individual observations (a, b).