An Automated Strategy for Binding-Pose Selection and Docking Assessment in Structure-Based Drug Design
An Automated Strategy for Binding-Pose Selection and Docking Assessment in Structure-Based Drug Design
复制标题
DOI：
10.1021/acs.jcim.5b00603
复制
发表时间：
2016-01-01
影响因子：
5.6
通讯作者：
Marshall, Garland R.
中科院分区：
文献类型：
--
作者：
Ballante, Flavio;Marshall, Garland R.
分子对接是药物设计中广泛使用的一种技术，用于预测候选化合物在确定的治疗靶点中的结合姿势。有许多对接方案可用，每一种都具有不同的搜索方法和评分功能，从而在同一配基蛋白质系统上提供不同的预测能力。为了验证对接协议，有必要先验地确定再现实验结合姿势的能力(即，通过确定对接精度(DA))，以便选择最合适的对接过程，从而估计对接新化合物的成功率。由于常见的对接程序通常使用不同的均方根偏差(RMSD)公式、计分函数和格式化结果，因此当在虚拟筛选期间在数千/数百万分子上应用给定对接程序时，一致地确定和比较它们的预测能力以识别用于感兴趣目标的最佳方案并推断结合姿势(即，最佳对接(BD)、最佳群集(BC)和最佳匹配(BF)姿势)是困难和耗时的。为了减少这一困难，已经开发并实现了两个名为Clusterizer和DockAccessor的新程序，以与一些常见的和“免费的”程序一起使用，例如AutoDock4、AutoDock4(Zn)、AutoDock Vina、Dock、MpSDockZn、Plants和Surflex-Dock，以自动推断BD、BC和BF姿势，并执行一致的簇和DA分析。Clusterizer和DockAccessor(互联网上可用的代码)代表了两个新的工具，用于收集通过计算确定的姿势并检测最具预测性的对接方法。本文介绍了人赖氨酸脱乙酰酶(HKDAC)抑制剂的应用。
Molecular docking is a widely used technique in drug design to predict the binding pose of a candidate compound in a defined therapeutic target. Numerous docking protocols are available, each characterized by different search methods and scoring functions, thus providing variable predictive capability on a same ligand protein system. To validate a docking protocol, it is necessary to determine a priori the ability to reproduce the experimental binding pose (i.e., by determining the docking accuracy (DA)) in order to select the most appropriate docking procedure and thus estimate the rate of success in docking novel compounds. As common docking programs use generally different root-mean-square deviation (RMSD) formulas, scoring functions, and format results, it is both difficult and time-consuming to consistently determine and compare their predictive capabilities in order to identify the best protocol to use for the target of interest and to extrapolate the binding poses (i.e., best-docked (BD), best-cluster (BC), and best-fit (BF) poses) when applying a given docking program over thousands/millions of molecules during virtual screening. To reduce this difficulty, two new procedures called Clusterizer and DockAccessor have been developed and implemented for use with some common and "free-for-academics" programs such as AutoDock4, AutoDock4(Zn), AutoDock Vina, DOCK, MpSDockZn, PLANTS, and Surflex-Dock to automatically extrapolate BD, BC, and BF poses as well as to perform consistent cluster and DA analyses. Clusterizer and DockAccessor (code available over the Internet) represent two novel tools to collect computationally determined poses and detect the most predictive docking approach. Herein an application to human lysine deacetylase (hKDAC) inhibitors is illustrated.