We present the first genetic study of depression including more than a million informative participants, with new large analyses from the Million Veteran Program meta-analyzed with prior results from the PGC + UK Biobank, 23andMe, and FinnGen, the largest analysis so far in what is a fast-moving field. We investigated genetic correlation between three different definitions (MDD-META, SR-Depression, and PHQ-2) of the depression phenotype within the MVP cohort. We identified 223 independently significant SNPs in 178 genomic loci associated with the primary meta-analysis, using an ICD code derived definition of depression for the MVP sample and GWAS summary statistics from 23andMe, UKB, PGC, and FinnGen. This finding is an increase of 77 loci over the largest previous study that investigated a comparable phenotype.10 As these cohorts used somewhat different definitions for depression (Table 1, Figure 1 Upper Left, Methods), we also used LDSC to examine genetic correlations between MVP depression phenotypes and these differentially defined depression phenotypes in independent cohorts. We investigated genetic correlation with 1,457 traits using available GWAS data, identifying 669 that were significantly correlated. We also used genomic structural equation modeling to evaluate how depression relates to other mental health and psychiatric phenotypes.
