Computational Toxicology Prioritization of CYP3A4 and DPP7 as Candidate Triclosan-Relevant Molecular Targets in Ulcerative Colitis.

Triclosan (TCS), a synthetic broad-spectrum antimicrobial classified as a novel persistent organic pollutant, is detected in over 75% of human urine samples, yet its molecular targets in ulcerative colitis (UC) remain uncharacterized at a systems level. We integrated network toxicology, multi-cohort machine learning, molecular docking, 100-ns all-atom molecular dynamics (MD) simulation, and single-cell transcriptomics to address this gap. Intersection of UC transcriptomic differentially expressed genes with database-derived TCS targets yielded 479 shared candidate targets; because these databases include indirect and probabilistic associations, this set was treated as a hypothesis-generating list for downstream feature compression rather than as evidence of direct molecular interference. KEGG enrichment identified the AGE-RAGE signaling pathway, PPAR signaling, and TNF signaling among the top enriched pathways. LASSO regression and random forest algorithms distilled a 10-gene diagnostic signature that achieved cross-cohort AUC of 0.910-1.000 across six independent GEO datasets, consistent with strong discriminative performance. Molecular docking on ligand-free receptors produced predicted Vina scores of -7.1 kcal/mol (CYP3A4) and -7.0 kcal/mol (DPP7); redocking recovered the co-crystal pose (0.85 Å RMSD for the TCS-sized DPP7 ligand) and small-molecule decoys scored markedly worse, supporting protocol specificity but not biochemical binding or inhibition. Over 100 ns of MD simulation, the docked poses remained stable for both targets. Single-cell RNA sequencing of GSE214695 independently corroborated cell-type-specific CYP3A4 downregulation in intestinal epithelial cells and broad DPP7 upregulation across immune subsets in UC. In silico perturbation via CellOracle predicted that CYP3A4 suppression would impair cytoplasmic translation, while DPP7 suppression would polarize innate inflammation toward an IL-1β-driven phenotype. Together, these analyses computationally prioritize CYP3A4 and DPP7 as candidate TCS-relevant targets in UC and generate testable hypotheses for future experimental validation; they do not by themselves establish direct target engagement, inhibition, or causal contribution to UC.