We applied MiXeR45 on our depression GWAS summary statistics and a selection of additional traits (Supplementary Table S6A, Figure 2 and Supplementary Figure 9-1) to estimate (i) the number of variants explaining 90% of the SNP heritability of each trait and (ii) the genetic overlap between depression and each trait. The additional traits included in the analyses are ADHD46, anxiety (FinnGen22 + MVP112), autism15, bipolar disorder118, educational attainment119, Neuroticism120, schizophrenia121, Smoking Initiation117 and substance use disorder (SUD)39,115. For comparison we included non-psychiatric traits, i.e. Alzheimer122, epilepsy123 and height124 not necessarily expected to show strong genetic correlations with depression. MiXeR analysis were conducted with default settings (https://github.com/precimed/mixer) in a two-step process: 1) a univariate model for each trait to produce estimates of the proportion of variants with non-zero additive genetic effect on the trait (i.e. “polygenicity”) and the variance of effect sizes of these non-zero variants (i.e. “discoverability”’). 2) the variance estimates obtained in the univariate analysis were applied in the bivariate model (i.e. depression vs. each of the additional traits) to obtain four estimates representing (i) zero-effect SNPs in both
