Elucidating the molecular mechanisms linking bisphenol A to breast cancer: an integrated study of bioinformatics, machine learning, and molecular docking.

Bisphenol A (BPA) is a prevalent environmental endocrine disruptor linked to breast cancer. However, the precise molecular mechanisms and core therapeutic targets remain to be fully elucidated. This study employed an integrative multi-omics approach to explore the potential mechanism of BPA-associated breast cancer. We integrated multiple transcriptomic datasets from the Gene Expression Omnibus (GEO) database and identified intersection targets between BPA and breast cancer through differential expression analysis, WGCNA, and multi-source database predictions (ChEMBL, PharmMapper, SEA). Pathway enrichment analyses revealed that these targets are predominantly involved in key signaling cascades, such as MAPK and PI3K/Akt. To identify robust biomarkers, we constructed a diagnostic model using machine learning algorithms and prioritized core genes via SHapley Additive exPlanations (SHAP) value analysis. Five core genes (EGFR, PPARG, MMP12, ADRB2, and KIF11) were identified, all of which demonstrated high diagnostic accuracy (AUC > 0.7) in validation sets. Subsequent molecular docking and molecular dynamics simulations predicted that BPA exhibits strong binding affinity (binding energy <  - 5 kcal/mol) to these core proteins. Collectively, our findings suggest that BPA may promote breast cancer progression by modulating these core targets and interfering with the MAPK/PI3K/Akt pathways. This study provides a data-driven theoretical basis for elucidating the molecular link between BPA and breast cancer, proposing potential biomarkers that warrant further investigation for clinical diagnosis and intervention.