To partition the heritability captured by top echelons of SNPs of a continuous annotation, we used the same strategy as in a previous study35. By this strategy, the proportion of heritability explained by a set of SNPs is the sum over all SNPs of the product of the τ* of each category in the S-LDSC model, for example, baseline-LD plus IMPACT annotation, and the SNP membership to that category (1 or 0 in the case of binary annotations, continuous values in the case of continuous annotations) divided by the same metric for all SNPs genome wide.