Improved Computation of Protein–Protein Relative Binding Energies with the Nwat-MMGBSA Method

A MMGBSA variant (here referred to as Nwat-MMGBSA), based on the inclusion of a certain number of explicit water molecules (Nwat) during the calculations, has been tested on a set of 20 protein–protein complexes, using the correlation between predicted and experimental binding energy as the evaluation metric. Besides the Nwat parameter, the effect of the force field, the molecular dynamics simulation length, and the implicit solvent model used in the MMGBSA analysis have been also evaluated. We found that considering 30 interfacial water molecules improved the correlation between predicted and experimental binding energies by up to 30%, compared to the standard approach. Moreover, the correlation resulted in being rather sensitive to the force field and, to a minor extent, to the implicit solvent model and to the length of the MD simulation.

本研究针对一种MMGBSA（分子力学-广义玻恩表面积法，Molecular Mechanics/Generalized Born Surface Area）变体展开测试，该变体在计算过程中引入了定量显式水分子（以Nwat指代显式水分子的数量），本文中将其记为Nwat-MMGBSA。测试数据集包含20组蛋白质-蛋白质复合物，以预测结合能与实验结合能之间的相关性作为评价指标。除Nwat参数外，本研究还考察了力场、分子动力学（MD）模拟时长以及MMGBSA分析中所采用的隐式溶剂模型的影响。结果发现，相较于标准分析方法，纳入30个界面水分子可将预测结合能与实验结合能的相关性提升最高达30%。此外，相关性结果对力场较为敏感，而对隐式溶剂模型与MD模拟时长的敏感性相对较弱。