Machine Learning and Molecular Modeling for Drug Repurposing Targeting Potential PI3Kα Inhibitors in Post-CoViD-19 Pulmonary Fibrosis.

Dysregulation of the phosphoinositide 3-kinase-alpha (PI3Kα) pathway is implicated in the development of post-CoViD-19 pulmonary fibrosis, highlighting the need for effective therapeutic agents. This study aimed to identify novel PI3Kα inhibitors by computationally repurposing FDA-approved drugs. We employed a hybrid approach that combines machine learning with molecular modeling. A random forest (RF) classification model was built and validated using a curated data set of 4,023 known PI3Kα inhibitors from the ChEMBL database, demonstrating robust predictive performance. The RF model was applied to screen the subset of FDA-approved drugs available in the DrugBank database to identify potential candidates. The top-ranked compounds were subsequently evaluated through molecular docking, extensive 200 ns molecular dynamics simulations (MDS), and binding free energy calculations using the molecular mechanics/Poisson-Boltzmann surface area (MM/PBSA) method. Our virtual screening identified five promising drugs, with simeprevir and ceritinib demonstrating the most favorable free energy binding affinities (Δ