We used two primary methods, partitioned LD Score regression (pLSDC)75 and MAGMA (v1.08),45 to test the enrichment of tissues and cell types in the MDD GWAS results. Our analyses using pLDSC evaluated if the SNPs within 100kb regions of the top 10% specifically expressed genes were enriched for SNP-based heritability. For each tissue or cell-type, we computed the LD scores for this cell-type-specific annotation and added it to the baseline model of 53 functional annotations. We assessed the enrichment of tissue or cell-types using the coefficient z-scores and computed one-sided p-values. For the analyses using MAGMA, we tested if the top 10% specifically expressed genes in each tissue or cell-type were the most associated genes from the GWAS. As part of this analysis, we filtered SNPs with MAF <1% or with imputation INFO score < 0.9, and mapped SNPs to genes with 35kb upstream and 10kb downstream windows. We first conducted gene-level association tests and then gene-set analyses for the tissue or cell-type specifically expressed genes. For both methods, we used the European samples in the phase 3 of 1000 Genomes Project as the reference panel and reported significance at the 5% false discovery rate within each dataset and method.
