Abstract
<title>Abstract</title> <p>Genetic risk for major depressive disorder (MDD) can act through gene expression or pre-mRNA alternative splicing, but effector-gene mapping has relied mainly on expression quantitative trait loci (eQTLs), leaving splicing-level effects less systematically resolved. We integrated the latest MDD GWAS (MDD2025; effective sample size ~ 1.15×10⁶) with BigBrain, a human-brain cis-splicing-QTL (sQTL) resource of 10 725 RNA-seq samples. Within risk regions, colocalization identified 56 high-confidence brain splicing-effector genes (PP.H4 > 0.8; median 0.92), compared with approximately 22 using the 2018 GWAS. A matched BigBrain eQTL analysis separated expression-orthogonal associations from limited eQTL power: 29 genes had a genome-wide-significant cis-eQTL at the same locus that did not colocalize with MDD, defining an expression-orthogonal core; 23 had no genome-wide-significant cis-eQTL in the corresponding locus, and only four showed eQTL–MDD colocalization, with a concordant pattern in MetaBrain. Splicing Mendelian randomization supported a consistent cis-sQTL-to-MDD direction. Across disorders, 35 of 56 genes were shared with other psychiatric conditions, whereas only one colocalized with Alzheimer’s or Parkinson’s disease. Seventeen genes were small-molecule tractable, including CACNA1C, targeted by approved calcium-channel blockers, and HDAC3, engaged by approved non-selective HDAC inhibitors. Thirteen genes remained high-confidence under a conservative prior, and 38 of 56 loci were single-signal. In an exploratory analysis, splicing-effector genes showed higher AlphaGenome-predicted splicing disruption than background genes (62% vs 45%; one-sided P = 0.03). These results define MDD candidate genes whose associations are more clearly resolved through brain splicing than expression in current bulk-brain resources, thereby providing a complementary regulatory layer for functional and translational prioritization.</p>