Machine learning assisted in Silico discovery and optimization of small molecule inhibitors targeting the Nipah virus glycoprotein

The Nipah virus (NiV), a lethal pathogen from the Paramyxoviridae family, presents a significant global health threat as a result of its high mortality rate and inter-human transmission. This investigation employed in silico methods that were assisted by machine learning to identify small-molecule inhibitors that target the NiV glycoprotein, a critical component of viral entry. Out of the 754 antiviral compounds that were screened using Lipinski's Rule of Five and DeepPurpose, 333 are identified. Five best hits were identified through molecular docking, each of which exhibited superior binding scores in comparison to the control. This was further refined to three compounds through density functional theory (DFT) analysis, with compound 138,567,123 exhibiting the highest electronic stability (DFT energy: -1976.74 Hartree; HOMO-LUMO gap: 0.83 eV). Its stability was verified by molecular dynamics (MD) simulations, which demonstrated consistent hydrogen bonding and minimal RMSD. Additionally, it possessed the highest docking score (-9.7 kcal/mol) and binding free energy (-24.04 kcal/mol, MM/GBSA). The results underscore ligand 138,567,123 as a promising antiviral candidate for NiV and illustrate the efficacy of machine learning-based in silico drug discovery.