Structural and energetic analysis of NS5 protein inhibition by small molecules in Japanese encephalitis virus using machine learning and steered molecular dynamics approach.

One of the viral diseases that affect millions of people around the world, particularly in developing countries, is (JE). In this study, the conserved protein of this virus, that is, non-structural protein 5 (NS5), was used as a target protein for this study, and a compound library of 749 antiviral molecules was screened against NS5. The current study employed machine learning-based virtual screening combined with molecular docking. Here, three hits (, and ) had lower binding energies (< -8 kcal/mol) than the control, S-Adenosyl-L-homocysteine (SAH). All the compounds showed significant H-bond interactions with functional residues, which were also observed by the control. Molecular dynamics simulation, MM/GBSA for binding free energy analysis, principal component analysis and free energy landscape were also performed to study the stability of the complex formation. All three compounds had similar root mean square deviation trends, which were comparable to the control, SAH. Post-MD, the -receptor complex had the highest number of H-bonds (4 to 5) after the control, out of which three exhibited the highest percentage occupancy (50%, 24% and 79%). Both docking and MD, showed an H-bond with the residue Gly111, which was also found for the control-protein complex. showed the lowest binding free energy with Δ of -89 kJ/mol. Steered molecular dynamics depicted that had the maximum magnitude of dissociation (1436.43 kJ/mol/nm), which was more than the control, validating its stable complex formation. This study concluded that is a binder and could inhibit the protein NS5 of JE.Communicated by Ramaswamy H. Sarma.