Fig. 5: Sample size is critical for detecting concordant effects between males and females.

The percentage of all UK Biobank main effect loci for BMI with consistent direction of effects in the two sexes was plotted according to the male-derived p values. Samples were randomly downsampled to simulate smaller sample sizes. Main effect p values > 0.95 were excluded due to many low-frequency alleles. Directionality was determined using the sign of the effect size of each main effect loci.