Xu Mingyuan, Zhu Tong, Zhang John Z H
State Key Lab of Precision Spectroscopy, Shanghai Engineering Research Center of Molecular Therapeutics & New Drug Development, Shanghai Key Laboratory of Green Chemistry & Chemical Process, School of Chemistry and Molecular Engineering , East China Normal University , Shanghai 200062 , China.
NYU-ECNU Center for Computational Chemistry at NYU Shanghai , Shanghai 200062 , China.
J Phys Chem A. 2019 Aug 1;123(30):6587-6595. doi: 10.1021/acs.jpca.9b04087. Epub 2019 Jul 24.
An artificial neural network provides the possibility to develop molecular potentials with both the efficiency of the classical molecular mechanics and the accuracy of the quantum chemical methods. Here, we develop an ab initio based neural network potential (NN/MM-RESP) to perform molecular dynamics study of zinc ion in liquid water. In this approach, the interaction energy, atomic forces, and atomic charges of zinc ion and water molecules in the first solvent shell are described by a neural network potential trained with energies and forces generated from density functional calculations. The predicted energies and forces from the NN potential show excellent agreement with the quantum chemistry calculations. Using this approach, we carried out molecular dynamics simulation to study the hydration of zinc ion in water. The experimentally observed zinc-water radial distribution function, as well as the X-ray absorption near edge structure spectrum, is well-reproduced by the MD simulation. Comparison of the results with other theoretical calculations is provided, and important features of the present approach are discussed. The neural network approach used in this study can be applied to construct potentials to study solvation of other metal ions, and its salient features can shed light on the development of more accurate molecular potentials for metal ions in other environments such as proteins.
人工神经网络为开发兼具经典分子力学效率和量子化学方法准确性的分子势提供了可能性。在此，我们开发了一种基于从头算的神经网络势（NN/MM-RESP），用于对液态水中的锌离子进行分子动力学研究。在这种方法中，通过用密度泛函计算产生的能量和力训练的神经网络势来描述第一溶剂层中锌离子与水分子的相互作用能、原子力和原子电荷。神经网络势预测的能量和力与量子化学计算结果显示出极好的一致性。利用这种方法，我们进行了分子动力学模拟以研究锌离子在水中的水合作用。MD模拟很好地再现了实验观察到的锌 - 水径向分布函数以及X射线吸收近边结构光谱。提供了与其他理论计算结果的比较，并讨论了本方法的重要特征。本研究中使用的神经网络方法可用于构建势来研究其他金属离子的溶剂化作用，其显著特征可为开发更精确的、用于诸如蛋白质等其他环境中金属离子的分子势提供启示。
