DOX_BDW: Incorporating Solvation and Desolvation Effects of Cavity Water into Nonfitting Protein-Ligand Binding Affinity Prediction

作者：Liu, Jiaqi; Wan, Jian^*; Ren, Yanliang; Shao, Xubo; Xu, Xin^*; Rao, Li^*

来源：Journal of Chemical Information and Modeling, 2023, 63(15): 4850-4863.

DOI：10.1021/acs.jcim.3c00776

• MOLECULAR-DYNAMICS SIMULATIONS
• EXTENDED ONIOM METHOD
• FREE-ENERGY
• COVALENT DOCKING
• SCORING FUNCTION
• QUANTUM-MECHANICS
• DRUG DISCOVERY
• SITE WATERS
• SOLVENT
• HYDRATION

Accurate prediction of the protein-ligand bindingaffinity(PLBA) with an affordable cost is one of the ultimate goals in thefield of structure-based drug design (SBDD), as well as a great challengein the computational and theoretical chemistry. Herein, we have systematicallyaddressed the complicated solvation and desolvation effects on thePLBA brought by the difference of the explicit water in the proteincavity before and after ligands bind to the protein-binding site.Based on the new solvation model, a nonfitting method at the first-principleslevel for the PLBA prediction was developed by taking the bridgingand displaced water (BDW) molecules into account simultaneously. Thenewly developed method, DOX_BDW, was validated against a total of765 noncovalent and covalent protein-ligand binding pairs,including the CASF2016 core set, Cov_2022 covalent binding testingset, and six testing sets for the hit and lead compound optimization(HLO) simulation. In all of the testing sets, the DOX_BDW method wasable to produce PLBA predictions that were strongly correlated withthe corresponding experimental data (R = 0.66-0.85).The overall performance of DOX_BDW is better than the current empiricalscoring functions that are heavily parameterized. DOX_BDW is particularlyoutstanding for the covalent binding situation, implying the needfor considering an electronic structure in covalent drug design. Furthermore,the method is especially recommended to be used in the HLO scenarioof SBDD, where hundreds of similar derivatives need to be screenedand refined. The computational cost of DOX_BDW is affordable, andits accuracy is remarkable.

• 单位
  复旦大学