We conducted polygenic profiling in three cohorts of African ancestry (48,669 cases and 52,939 controls), two cohorts of Latinx (AMR) ethnicity (1,202 cases and 5,112 controls), three cohorts of East Asian ancestry (6,902 cases and 75,879 controls), and three cohorts of South Asian ancestry (4,862 cases and 28,965 controls) and compared them to the 43 European cohorts. We included studies from China, the USA and the UK that had over 200 cases. Case status was defined based on symptom questionnaires, healthcare records or a combination of both (Table S7). The summary statistics used as a training set excluded 23andMe (because of data sharing restrictions). The available SNP set was limited to those with MAF > 0.10 and imputation INFO score > 0.9, and ambiguous SNPs were excluded. Further, we retained only SNPs with imputation INFO score > 0.9 in the target data sets.
