We used terminology and acronyms from the 1000 Genomes Project42 to describe ancestry. The majority of participating cohorts stratified their sample by ancestry before generation of summary statistics. Cohorts composed of substantial samples of multiple ancestry groups provided summary statistics stratified by ancestry, as well as results based on all individuals regardless of ancestry for use in the multi-ancestry meta-analyses. As TOPMed served multiple functions in the present study, including as an LD reference panel, we detailed the ancestry analyses and classification of TOPMed data in the Supplementary Note. For example, for both ancestry-stratified and multi-ancestry conditional analysis, we created TOPMed reference panels for estimating LD. We first created ancestry-stratified reference samples, resulting in matched ancestry reference sample sizes of n = 28,665 AFR, n = 19,737 AMR, n = 4,918 EAS and n = 51,656 EUR. To create a TOPMed-based reference sample for multi-ancestry analyses, we combined the matched ancestry individuals, resulting in a diverse ancestry reference panel (n = 104,976) that matches the ancestry proportions of the included cohorts to estimate LD.
