Hill Anthony D, Reilly Peter J
Department of Chemical and Biological Engineering, Iowa State University, Ames, Iowa 50011, USA.
J Comput Chem. 2008 May;29(7):1131-41. doi: 10.1002/jcc.20873.
Thermodynamic information can be inferred from static atomic configurations. To model the thermodynamics of carbohydrate binding to proteins accurately, a large binding data set has been assembled from the literature. The data set contains information from 262 unique protein-carbohydrate crystal structures for which experimental binding information is known. Hydrogen atoms were added to the structures and training conformations were generated with the automated docking program AutoDock 3.06, resulting in a training set of 225,920 all-atom conformations. In all, 288 formulations of the AutoDock 3.0 free energy model were trained against the data set, testing each of four alternate methods of computing the van der Waals, solvation, and hydrogen-bonding energetic components. The van der Waals parameters from AutoDock 1 produced the lowest errors, and an entropic model derived from statistical mechanics produced the only models with five physically and statistically significant coefficients. Eight models predict the Gibbs free energy of binding with an error of less than 40% of the error of any similar models previously published.
热力学信息可从静态原子构型中推断出来。为了准确模拟碳水化合物与蛋白质结合的热力学过程，已从文献中收集了一个大型结合数据集。该数据集包含来自262个独特蛋白质 - 碳水化合物晶体结构的信息，这些结构的实验结合信息是已知的。向这些结构中添加了氢原子，并使用自动对接程序AutoDock 3.06生成训练构象，从而得到了一个包含225,920个全原子构象的训练集。总共针对该数据集对AutoDock 3.0自由能模型的288种公式进行了训练，测试了计算范德华力、溶剂化和氢键能量成分的四种替代方法中的每一种。AutoDock 1的范德华参数产生的误差最小，并且从统计力学推导的熵模型产生了唯一具有五个物理和统计上显著系数的模型。八个模型预测结合的吉布斯自由能，其误差小于之前发表的任何类似模型误差的40%。
