Exploring Gingerol Substructures as Potential SARS‐CoV‐2 RBD Inhibitors: An Integrated Machine Learning and In Silico Approach

Abstract The ongoing COVID‐19 pandemic has underscored the urgent need for effective antiviral therapies. This study presents a novel approach that integrates machine learning (ML) and in silico techniques to identify potential inhibitors targeting the receptor‐binding domain (RBD) of SARS‐CoV‐2. A Random Forest (RF) regression model achieved an R 2 value of 0.82, accurately predicting Vina scores for 988 gingerol substructures. Subsequent virtual screening using PyRx identified 14 promising candidates with docking scores lower than that of the reference compound, arbidol. Pharmacokinetic analysis via the BOILED‐Egg model and Bioavailability Radar revealed four compounds‐ G4 , G5 , G11 , and G13 that exhibit favorable human intestinal absorption (HIA) and drug‐like properties, with G13 identified as a P‐glycoprotein (P‐gp) non‐substrate. Docking studies indicated significant interactions characterized by hydrogen bonding and hydrophobic contacts. Molecular dynamics (MD) simulations demonstrated the stability of the ligand‐protein complexes, showing a maximum root mean square deviation (RMSD) of 3 Å and low root mean square fluctuation (RMSF) values of 1.5 Å across interacting residues, highlighting the robustness of each ligand within the RBD binding pocket. Additionally, electrostatic potential (ESP) plots revealed highly electronegative regions around oxygen atoms, likely to engage in stabilizing interactions with positively charged or polar residues. Furthermore, binding stability assessments of G4 , G5 , G11 , and G13 against mutated RBDs of SARS‐CoV‐2, with docking scores ranging from −7.7 to −8.8 kcal mol −1 , indicate a strong affinity for the mutated protein, suggesting these compounds remain effective across various mutations. Our study provides a streamlined methodology for identifying potential inhibitors, paving the way for further experimental validation.