Lerner Michael G, Meagher Kristin L, Carlson Heather A
Department of Biophysics, University of Michigan, 930 North University Avenue, Ann Arbor, MI 48109-1055, USA.
J Comput Aided Mol Des. 2008 Oct;22(10):727-36. doi: 10.1007/s10822-008-9231-6. Epub 2008 Aug 5.
Use of solvent mapping, based on multiple-copy minimization (MCM) techniques, is common in structure-based drug discovery. The minima of small-molecule probes define locations for complementary interactions within a binding pocket. Here, we present improved methods for MCM. In particular, a Jarvis-Patrick (JP) method is outlined for grouping the final locations of minimized probes into physical clusters. This algorithm has been tested through a study of protein-protein interfaces, showing the process to be robust, deterministic, and fast in the mapping of protein "hot spots." Improvements in the initial placement of probe molecules are also described. A final application to HIV-1 protease shows how our automated technique can be used to partition data too complicated to analyze by hand. These new automated methods may be easily and quickly extended to other protein systems, and our clustering methodology may be readily incorporated into other clustering packages.
基于多拷贝最小化（MCM）技术的溶剂映射法在基于结构的药物发现中很常见。小分子探针的最小值定义了结合口袋内互补相互作用的位置。在此，我们提出了改进的MCM方法。特别是，概述了一种Jarvis-Patrick（JP）方法，用于将最小化探针的最终位置分组为物理簇。通过对蛋白质-蛋白质界面的研究测试了该算法，结果表明该过程在蛋白质“热点”映射中具有稳健性、确定性且速度快。还描述了探针分子初始放置的改进。对HIV-1蛋白酶的最终应用展示了我们的自动化技术如何用于对过于复杂而无法手动分析的数据进行划分。这些新的自动化方法可以轻松快速地扩展到其他蛋白质系统，并且我们的聚类方法可以很容易地纳入其他聚类软件包中。
