ChalcoVina: Enhancing Molecular Docking Accuracy by Incorporating Chalcogen Bond-Aware Scoring

Abstract

Molecular docking is a cornerstone technique in structure-based drug design (SBDD), providing a computationally efficient approach for predicting ligand–receptor interactions. However, conventional docking programs fail to accurately identify chalcogen bonds (ChBs), which play crucial roles in molecular recognition. Here, we present ChalcoVina, an augmented molecular docking program based on AutoDock Vina that simultaneously evaluates intermolecular and intramolecular ChBs for the first time. An intermolecular ChB term was incorporated into the Vina scoring function using 14,049 sets of quantum-mechanically derived geometric and energetic data. An empirical function constructed from prior knowledge and statistical data was further introduced to account for intramolecular ChBs. Docking experiments demonstrated that ChalcoVina improved docking success rates by 3.3%–13.1% and the geometric reproduction rate of intramolecular ChBs by approximately 23%, which significantly refined the prediction of binding poses. These enhancements expand the applicability of molecular docking for drug discovery involving sulfur-containing compounds. Furthermore, the approach for constructing and integrating ChB terms can be a blueprint for other σ-hole-mediated interactions, facilitating the inclusion of more nonclassical interactions in SBDD.