In addition, we performed a transcriptome-wide association study (TWAS), imputing the genetically regulated gene expression using EpiXcan49 and using models trained on PsychENCODE Consortium (PEC)50,51 expression data for genes and transcripts detected in the dorsolateral prefrontal cortex (DLPFC), with the aim of identifying and prioritizing putative causal loci for the broad depression phenotype definition. A total of 34,646 genes/transcripts were tested (transcripts with prediction performance R2>0.01 and prediction performance q-value <0.05 with the Benjamini-Hochberg method were retained), and applying a significance threshold of P<1.44x10−6 (corresponding to Bonferroni correction of all genes and isoforms tested; Figure 1D, Extended Data Figure 1, Supplementary Figure S10 and Table S9), using information on approximately independent linkage disequilibrium blocks in human populations52 to identify independent genomic regions with genes/transcripts showing significant differential gene expression between depression cases and controls.
