To identify IMPACT annotations enriched for causal genetic variation, we then partitioned the common SNP (minor allele frequency (MAF) > 5%) heritability of these 111 datasets using S-LDSC3 with an adapted baseline-LD model excluding cell-type-specific annotations31,35 (Supplementary Fig. 3 and Methods). We tested each of the traits against each of the 707 IMPACT annotations, assessing the significance of a nonzero τ*, which is defined as the proportionate change in per-SNP heritability associated with a 1 s.d. increase in the value of the annotation (Methods)35. Of 707 by 111 (n = 78,477) possible associations subjected to 5% false discovery rate (FDR), we detected 7,993 associations, where 95 phenotypes had at least one significant annotation association (τ* > 0, two-tailed z-test FDR < 0.05; Extended Data Fig. 2, Methods and Supplementary Tables 4–8). For narrative purposes, we exemplify our results using five genetically uncorrelated and biologically diverse traits, representative of the summary statistics analyzed. These five traits include an allergic phenotype (asthma), an autoimmune disease (rheumatoid arthritis (RA)), a neoplastic type (prostate cancer (PrCa)), a hematological quantitative trait (mean corpuscular volume (MCV))
