Integrative Machine Learning, Virtual Screening, and Molecular Modeling for BacA-Targeted Anti-Biofilm Drug Discovery Against Staphylococcal Infections

Abstract

The rise in antibiotic-resistant Staphylococcal infections necessitates innovative approaches to identify new therapeutic agents. This study investigates the application of machine learning models to identify potential phytochemical inhibitors against BacA, a target related to Staphylococcal infections. Active compounds were retrieved from BindingDB while the decoy was generated from DUDE. The RDKit was utilized for feature engineering. Machine learning models such as k-nearest neighbors (KNN), the support vector machine (SVM), random forest (RF), and naive Bayes (NB) were trained on an initial dataset consisting of 226 active chemicals and 2550 inert compounds. Accompanied by an MCC of 0.93 and an accuracy of 96%, the RF performed better. Utilizing the RF model, a library of 9000 phytochemicals was screened, identifying 300 potentially active compounds, of which 192 exhibited drug-like properties and were further analyzed through molecular docking studies. Molecular docking results identified Ergotamine, Withanolide E, and DOPPA as top inhibitors of the BacA protein, accompanied by interaction affinities of −8.8, −8.1, and −7.9 kcal/mol, respectively. Molecular dynamics (MD) was applied for 100 ns to these top hits to evaluate their stability and dynamic behavior. RMSD, RMSF, SASA, and Rg analyses showed that all complexes remained stable throughout the simulation period. Binding energy calculations using MMGBSA analysis revealed that the BacA_Withanolide E complex exhibited the most favorable binding energy profile with significant van der Waals interactions and a substantial reduction in gas-phase energy. It also revealed that van der Waals interactions contributed significantly to the binding stability of Withanolide E, while electrostatic interactions played a secondary role. The integration of machine learning models with molecular docking and MD simulations proved effective in identifying promising phytochemical inhibitors, with Withanolide E emerging as a potent candidate. These findings provide a pathway for developing new antibacterial agents against Staphylococcal infections, pending further experimental validation and optimization.