Integrative Machine Learning and Molecular Modeling-Based Discovery of Natural Inhibitors Targeting the HPV E7 Oncoprotein in Cervical Cancer

Cervical cancer, the fourth most common cancer in women globally, causes [Formula: see text]311,000 deaths annually. Among its various causes, the primary cause is human papillomavirus (HPV), a double-stranded DNA virus. The E7 oncoprotein promotes pRb degradation via ubiquitin-mediated proteasomal pathways, releasing E2F transcription factors that drive uncontrolled proliferation. It also activates the PI3K/AKT/mTOR signaling cascade to enhance cell survival. These molecular disruptions make E7 a key target for phytochemical-based therapeutic strategies. The current study employed machine learning-based virtual screening to identify novel phytochemical inhibitors against E7. A total of 2,301 medicinal plant-derived compounds from the MPD3 library were screened using a highly accurate random forest model (with 99% accuracy), resulting in the screening of 144 drug-like compounds against the E7. These hits were subjected to molecular docking to evaluate their binding affinities with the E7 oncoprotein. Based on docking scores, three top-ranked compounds were identified: PubChem 3985862 (−4 kcal/mol), PubChem 6951175 (−4.2 kcal/mol), and PubChem 3914876 (−4 kcal/mol). These were further analyzed through ADMET profiling to assess their pharmacokinetic properties. To evaluate their stability and dynamic behavior, 100 ns molecular dynamics (MD) simulations were performed. Among the three, PubChem 6951175 demonstrated the most stable binding throughout the simulation, with the lowest MM-GBSA binding energy (−9.28 kcal/mol), indicating a strong interaction with the E7 protein. Overall, PubChem 6951175 might act as a potential inhibitor of the E7 oncoprotein. These findings also suggest that combining machine learning with molecular modeling techniques could help streamline the identification of targeted candidates for cervical cancer therapy.