Liu Haiyan, Lu Zhenyu, Cisneros G Andres, Yang Weitao
School of Life Science, University of Science and Technology of China, Hefei, Anhui, 230026, China.
J Chem Phys. 2004 Jul 8;121(2):697-706. doi: 10.1063/1.1759318.
The determination of reaction paths for enzyme systems remains a great challenge for current computational methods. In this paper we present an efficient method for the determination of minimum energy reaction paths with the ab initio quantum mechanical/molecular mechanical approach. Our method is based on an adaptation of the path optimization procedure by Ayala and Schlegel for small molecules in gas phase, the iterative quantum mechanical/molecular mechanical (QM/MM) optimization method developed earlier in our laboratory and the introduction of a new metric defining the distance between different structures in the configuration space. In this method we represent the reaction path by a discrete set of structures. For each structure we partition the atoms into a core set that usually includes the QM subsystem and an environment set that usually includes the MM subsystem. These two sets are optimized iteratively: the core set is optimized to approximate the reaction path while the environment set is optimized to the corresponding energy minimum. In the optimization of the core set of atoms for the reaction path, we introduce a new metric to define the distances between the points on the reaction path, which excludes the soft degrees of freedom from the environment set and includes extra weights on coordinates describing chemical changes. Because the reaction path is represented by discrete structures and the optimization for each can be performed individually with very limited coupling, our method can be executed in a natural and efficient parallelization, with each processor handling one of the structures. We demonstrate the applicability and efficiency of our method by testing it on two systems previously studied by our group, triosephosphate isomerase and 4-oxalocrotonate tautomerase. In both cases the minimum energy paths for both enzymes agree with the previously reported paths.
确定酶系统的反应路径仍然是当前计算方法面临的巨大挑战。在本文中，我们提出了一种使用从头算量子力学/分子力学方法确定最小能量反应路径的有效方法。我们的方法基于对Ayala和Schlegel用于气相小分子的路径优化程序的改编、我们实验室早期开发的迭代量子力学/分子力学（QM/MM）优化方法以及引入一种定义构型空间中不同结构之间距离的新度量。在该方法中，我们用一组离散的结构来表示反应路径。对于每个结构，我们将原子划分为通常包括量子力学子系统的核心集和通常包括分子力学子系统的环境集。这两组进行迭代优化：核心集被优化以近似反应路径，而环境集被优化到相应的能量最小值。在优化反应路径的原子核心集时，我们引入一种新度量来定义反应路径上各点之间的距离，该度量排除了环境集中的软自由度，并在描述化学变化的坐标上包含额外权重。由于反应路径由离散结构表示，并且每个结构的优化可以在非常有限的耦合下单独进行，我们的方法可以自然且高效地并行执行，每个处理器处理其中一个结构。我们通过在我们小组之前研究过的两个系统——磷酸丙糖异构酶和4-草酰巴豆酸互变异构酶上进行测试，证明了我们方法的适用性和效率。在这两种情况下，两种酶的最小能量路径都与先前报道的路径一致。
