Supplementary Table 2 Ascertainment and standard errors in simulations at different sample sizes.

For each annotation, simulation runs were discarded if heritability for that annotation was not significantly different from zero (Z < 2) or if the standard error of the M_a estimate for the annotation was larger than two times the median M_a point estimate. For each annotation, the fraction of simulation runs that were ascertained is reported, as well as the standard deviation of the normalized residuals for the ascertained simulations. The normalized residuals are defined as divided by the jackknife standard error, and they should be equal to one if the jackknife standard errors are perfectly calibrated. Values less than 1 indicate conservative standard errors. Jackknife standard errors for polygenicity enrichment are nearly proportional to standard errors for Ma, since there is relatively little noise in the denominator (the estimate of M_a for all common and LF SNPs).