online functions for trait causality investigation, there are still some points that can be further improved in future. Many complex genetic loci harbor multiple causal variants for a particular trait/disease (54,55); given the computational burden and relatively low accuracy of multi-causal variants inference, CAUSALdb only assumes a single causal signal in each independent LD block. However, it allows users to download block-wise summary statistics for customized fine-mapping. In addition, several fine-mapping approaches in trans-ethnics have been proposed (56); CAUSALdb excludes GWASs with mixed populations in the current version because of complex LD patterns. As the GWAS summary data in CAUSALdb are from worldwide populations and considering the restrictions on obtaining original individual genotypes or large LD references (57), we only extracted LD information from 1KGP, which may not reflect the actual LD pattern in the corresponding GWAS cohort. In future studies, we plan to address the aforementioned issues by adding new features to CAUSALdb. We also aim to perform monthly curation of newly available GWAS summary statistics and frequently update variant information and functional annotations. In conclusion, with the accumulation of summary-level GWAS data, we believe that CAUSALdb should considerably aid researchers to interrogate the genetic mechanisms underlying diseases, thereby
