Fig. 1: Characteristics of functional annotation data.

a, Functional annotations provide orthogonal information that helps to distinguish the causal variant (CV) from the SNP in perfect LD with it. However, when the causal variant is not observed, its effect can be captured through LD by an SNP that has a different annotation from the causal variant, resulting in a mismatch between effect size and annotation category (denoted by ‘Annot’). b, Functional categories can differ in both the proportion of causal variants and the distribution of causal effect sizes, either of which can lead to an enrichment or depletion in per-SNP heritability in a functional category.