Fig. 2

Convergence of common and rare genetic effects as a function of sample size. The CORAC statistic (y-axis) is positively correlated with effective sample size (x-axis). The positive relationship remained after accounting for heteroskedasticity (P < 2.2e − 16). The significant relationship was also observed after further adjustment for the number of detected association signals and after sensitivity analysis that varied the number of significant genes (20, 50, and 200 top genes). The regression line and the 95% confidence bands are shown, allowing identification of traits with a higher or lower convergence signature than expected given the effective sample size. Traits with relatively high and low CORAC values are colored blue and orange, respectively. Two traits (body mass index and standing height) located close to the regression line are colored grey (a). We implemented SbayesS and estimated the degree of polygenicity (pi) for each trait. Traits (colored orange in a) with low CORAC values show a significantly higher degree of polygenicity than traits (colored blue in a) with high CORAC values (P = 3.5e − 4, b)