Department of Medicinal Chemistry, College of Pharmacy, University of Michigan, 428 Church Street, Ann Arbor, Michigan 48109-1065, USA.
J Chem Inf Model. 2013 Feb 25;53(2):391-402. doi: 10.1021/ci300430v. Epub 2013 Feb 13.
Computational approaches to fragment-based drug design (FBDD) can complement experiments and facilitate the identification of potential hot spots along the protein surface. However, the evaluation of computational methods for mapping binding sites frequently focuses upon the ability to reproduce crystallographic coordinates to within a low RMSD threshold. This dependency on the deposited coordinate data overlooks the original electron density from the experiment, thus techniques may be developed based upon subjective-or even erroneous-atomic coordinates. This can become a significant drawback in applications to systems where the location of hot spots is unknown. On the basis of comparison to crystallographic density, we previously showed that mixed-solvent molecular dynamics (MixMD) accurately identifies the active site for HEWL, with acetonitrile as an organic solvent. Here, we concentrated on the influence of protic solvent on simulation and refined the optimal MixMD approach for extrapolation of the method to systems without established sites. Our results establish an accurate approach for comparing simulations to experiment. We have outlined the most efficient strategy for MixMD, based on simulation length and number of runs. The development outlined here makes MixMD a robust method which should prove useful across a broad range of target structures. Lastly, our results with MixMD match experimental data so well that consistency between simulations and density may be a useful way to aid the identification of probes vs waters during the refinement of future multiple solvent crystallographic structures.
基于片段的药物设计（FBDD）的计算方法可以补充实验，并有助于确定蛋白质表面的潜在热点。然而，评估用于映射结合位点的计算方法的频率集中在能够将晶体学坐标复制到低 RMSD 阈值内的能力上。这种对已存入坐标数据的依赖性忽略了实验中的原始电子密度，因此可能会基于主观甚至错误的原子坐标来开发技术。在应用于热点位置未知的系统时，这可能成为一个重大缺陷。基于与晶体学密度的比较，我们之前表明，混合溶剂分子动力学（MixMD）可以准确识别 HEWL 的活性位点，其中乙腈作为有机溶剂。在这里，我们集中研究质子溶剂对模拟的影响，并改进了最佳的 MixMD 方法，以便将该方法外推到没有确定位置的系统。我们的结果为比较模拟与实验建立了一种准确的方法。我们已经根据模拟长度和运行次数概述了 MixMD 的最有效策略。这里概述的发展使 MixMD 成为一种稳健的方法，应该对广泛的目标结构都有用。最后，我们使用 MixMD 的结果与实验数据非常吻合，因此模拟和密度之间的一致性可能是在未来的多溶剂晶体结构精修过程中辅助识别探针与水的一种有用方法。
