Investigation of Protein-Ligand Interaction Using an Intermolecular Distance-Based Descriptor in Molecular Docking

Molecular docking is a key tool in structure-based drug design, extensively used to study biomolecular interactions and mechanisms. Molecular docking reliability is often evaluated using RMSD (Root Mean Square Deviation) compared to experimental structures, though such data is frequently unavailable in practice. Therefore, scoring functions can be used as an alternative to assess protein-ligand docking results. In this study, a simple computational scoring function for protein-ligand interaction was developed, based on calculating the sum of pairwise distances between ligand atoms bound in the active site and protein atoms. The distance matrix can be used to calculate a distance-based score (DB-Score). To better evaluate performance, we used experimentally determined values for IC_۵۰ GRK۶ (G protein-coupled receptor kinase ۶) inhibitors to assess scoring and ranking accuracy compared to the AutoDock Vina program performances. Extensive experiments on this dataset demonstrate that the distance-based scoring function outperforms the conventional AutoDock Vina score in ranking and scoring. Pearson’s correlation coefficients for AutoDock Vina and our defined score against experimentally determined GRK۶-pIC_۵۰ were ۰.۰۹ and ۰.۷۶, respectively. Furthermore, the effectiveness of DB-Score was evaluated using the v۲۰۱۶-core subset of the PDBbind database. On the CASF-۲۰۱۶ benchmark, DB-Score achieved a Pearson’s r of ۰.۶۲, demonstrating surprisingly good performance.

کلیدواژه ها:

Molecular docking ، distance-based score ، structure-based drug design ، protein-ligand interaction ، scoring function