Many cohorts cannot share individual phenotype and genotype data but can contribute case-control GWAS summary statistics. We incorporated summary statistics from 24 independent cohorts of European ancestry (496,710 cases and 3,010,973 controls) into the meta-analysis. Building on Meng et al.,9 we also incorporated data from ancestrally diverse cohorts, including 8 cohorts with participants of African ancestry (9,649 cases and 122,347 controls), 7 with East Asian ancestry (18,709 cases and 349,619 controls), 1 with South Asian ancestry (3,748 cases and 25,934 controls), 5 with Hispanic/Latin American ethnicity (19,927 cases and 340,403 controls). For the first time our analyses accommodated cohorts with participants of diverse and mixed ancestry by using a joint analysis approach (12 cohorts, 108,578 cases and 163,587 controls). We excluded participants of European ancestry from these studies where they had already been included in the analyses described above. The numbers of cases and controls, and MDD assessment methods are summarized in Table S1. Additional information, including genotyping, quality control and imputation are described in the Methods S1. Methods for determining MD status included clinical interviews, health register or medical records, self-reported questionnaires, and self-report of diagnosis.
