Prioritization of novel druggable targets for epilepsy via Mendelian randomization and colocalization

Epilepsy is clinically heterogeneous and frequently drug-resistant, motivating genetics-informed target prioritization with clinical tractability. Two-sample Mendelian randomization using whole-blood cis-eQTL instruments estimated associations between genetically proxied gene expression and the odds of epilepsy in FinnGen R11 (14,089 cases). For MR-prioritized genes, Bayesian colocalization assessed whether eQTL and GWAS signals were statistically consistent with a shared causal variant at each locus. Biological plausibility and potential safety liabilities were contextualized via protein-protein interaction analysis, functional enrichment, and phenome-wide association screening. ADORA1 and TKT met the prespecified Tier 1 colocalization criterion (PPH4 ≥ 0.90), indicating high posterior probability consistent with a shared causal variant at their respective loci. EP300, CXCR6, and CLEC3B showed suggestive colocalization (0.80 ≤ PPH4 < 0.90) and were retained as secondary, hypothesis-generating signals. Functionally, prioritized genes map to pathways related to adenosine signaling, epigenetic regulation, immune-vascular biology, and metabolism. For pharmacological contextualization, compound-gene signature resources were queried, with triage excluding agents with established severe toxicity, carcinogenicity, or environmental pollutant status and annotation of remaining hits by regulatory status. Molecular docking was used as a structural plausibility screen; docking scores were not interpreted as evidence of in vivo efficacy, safety, or clinical druggability. Inference is constrained by blood-derived regulatory instruments and outcome evidence from a single GWAS cohort; brain-relevant regulation and generalizability require replication. This study provides a transparent genetics-informed framework and a tiered candidate set for discovery-stage mechanistic validation in epilepsy.