Antipsychiatric Pharmacogenomics Validation Benchmark

This dataset is purpose-built to stress-test pharmacogenomic interpretation engines in clinical and research environments. It covers 16 drugs across four major psychiatric classes including SSRIs, TCAs, atypical antipsychotics, and atomoxetine—with 23,000 synthetic patients and over 35,000 gene-level rows. Unlike static genotype-only sets, this benchmark models real-world metabolic complexity through a three-factor phenoconversion layer: drug-drug-gene interactions (DDGI), smoking-induced CYP1A2 induction, and inflammation (CRP)-mediated CYP suppression. Genotypes are ancestry-stratified using published allele frequencies across European, African, Asian, and Latin American populations. Every record includes a pre-calculated, source-cited ground truth, enabling automated, deterministic validation of your AI's outputs. Ideal for ML training, regulatory submission support, health-equity auditing, and regression testing of clinical decision-support systems.