The genomic distribution of these variants showed that 98.9% of them were located in non-coding genomic regions (Supplementary Figure S5), emphasizing the pivotal role of regulatory variants in the development of complex traits. Among these variants, ∼55% were identified by more than one study, which implies that shared genetic effects among traits could be very common. Notably, many test variants with genome-wide significance (P-value ≤ 5E−8) were not present in the credible set, which included 1 339 760 unique variants and 11 528 369 genotype-phenotype associations, demonstrating that fine-mapping can greatly narrow down potential causal hits. In 1703 relatively independent LD blocks of EUR population defined by LDetect (25), 1699 blocks were noted to contain potential causal variants. Among these blocks, 99.88% contained potential causal variants across multiple studies; 6p21.32 (HLA locus) was the top causal block, associated with 514 studies and 176 traits. To investigate the potential pleiotropic variants in CAUSALdb, we performed joint analysis in GWAS pairs using gwas-pw (29). We in total identified 24 286 potential variant-level pleiotropy (PP > 0.5 in gwas-pw model 3) for 219 880 GWAS pairs. Although this estimation cannot further distinguish vertical or horizontal pleiotropy, it may help users cautiously interpret the
