the downstream analyses using MatrixQTL. The gene expression was corrected for the batch, age, sex, RNA integrity number (RIN), Post-mortem Interval (PMI), and alcohol status using the “removeBatchEffect” option from the limma package. The genetically derived PC1 was also added as additional covariate in the eQTL analysis. The eQTL summary statistics from PsychEncode49, ROSMAP50, and COGA-INIA datasets were processed and munged together at single bp and allele level to remove ambiguity due to dbSNP rsids. The gene labels in all three datasets were also matched to Ensembl ids. The summary statistics were saved in binary format files (BESD) using the SMR (https://cnsgenomics.com/software/smr/#DataManagement). The meta-analysis for eQTLs was performed using the conventional inverse-variance-weightedmeta-analysis assuming all cohorts are independent. SMR “–meta” option was used to perform the meta-analysis in all three datasets.
