To perform fine-mapping on curated GWAS summary statistics, we partitioned the genetic variants with relatively independent LD blocks estimated using LDetect (25). We checked each file and extracted the variants in LD blocks (termed causal blocks) that had at least one genome-wide significant variant (P-value ≤ 5E−8). For studies without any genome-wide significant variants (P-value ≤ 5E−8), we only selected the LD block in which the variant with the lowest genome-wide P-value located as the potentially causal block for each trait. The GWAS summary information for each causal block was then reformatted into the format required by the fine-mapping tools.
