Integrating machine learning and structural analysis to decipher benzo[a]pyrene-induced bladder cancer networks

Benzo[a]pyrene (BaP), a polycyclic aromatic hydrocarbon from tobacco smoke, exhaust, and pollutants, is linked to bladder cancer (BLCA). We systematically analyzed GEO datasets to identify BaP-related differentially expressed genes (DEBRGs). By integrating network toxicology, machine learning, molecular docking, molecular dynamics, and single-cell transcriptomics, we identified 19 significant genes, among which 7 key DEBRGs were prioritized (GSK3B, SKP2, AURKB, EPHB4, KIT, NR3C2, and CA2) in BaP-mediated BLCA. SHAP analysis highlighted GSK3B and SKP2 as important genes contributing to the predictive model. Single-cell data revealed their cell-type specific expression in the tumor microenvironment. Molecular simulations detailed interactions between BaP and target proteins. This study identified critical genes in BaP-induced bladder carcinogenesis, offering insights into underlying molecular mechanisms and potential therapeutic targets.