We started with the collection and curation of GWAS summary statistics from various resources and publications (details in Materials and Methods). After pre-fine-mapping QC, up to the latest update in July 2019, CAUSALdb curated 3052 fine-mappable GWAS summary statistics in total: 1237 belonged to non-UKBB cohorts and 1815 belonged to the UKBB cohort. In total, 2629 unique traits were identified that could be mapped to 855 MeSH terms. According to the ontology mapping, around two-thirds of the studies were based on common diseases such as cardiovascular diseases and neoplasms, while the remainder focused on quantitative traits of human phenotypes (Supplementary Figure S1). In the non-UKBB cohort data, 92.07% of studies were based on the EUR population, 6.91% on EAS population, and only seven, six, and two studies were based on AMR, AFR and SAS populations, respectively (Supplementary Figure S2), which indicates an unequal ancestry composition in the current GWASs. The average sample size in the non-UKBB cohort studies was 43 516, with a meta-analysis of atrial fibrillation (48) having the largest sample size (1 030 836). In the UKBB cohort
