For post-GWAS analysis, FinnGen was removed a priori due to potential for increased heterogeneity in the phenotype definition due to the broad nature of inclusion in the F5 Mood phenotype. Genetic correlation analyses were performed using LDSC to assess the degree of genetic overlap between phenotypes and across the cohorts included in the analysis. Per-trait observed-scale SNP-based heritability estimates were calculated via LDSC using the 1000 Genomes Project European linkage disequilibrium reference panel. 13 Heritability estimates were calculated for 1,468 phenotypes from FinnGen, 4,083 phenotypes from UKB, 3,143 brain image derived phenotypes from the Oxford Brain Imaging Genetics (BIG) project, and phenotypes from the Psychiatric Genomics Consortium (PGC), the Social Science Genetic Association Consortium (SSGAC), and the Genetics of Personality Consortium (GPC). Heritability z-scores were calculated by dividing the heritability estimate per phenotype by its associated standard error. Phenotypes with heritability z-scores ≥ 4 were considered suitable for genetic correlation against MDD-META. 13 For continuous UKB phenotypes we restricted our analyses to use inverse-rank normalized phenotypes instead of untransformed phenotypes. Genetic correlations are summarized by total phenotypes tested, nominally significant (p<0.05), and after application of 5% false discovery rate and Bonferroni thresholds (Figure 2 Lower Panel).
