Laboratory of Physical Chemistry, Swiss Federal Institute of Technology ETH, 8093 Zürich, Switzerland.
J Chem Phys. 2012 Feb 7;136(5):054505. doi: 10.1063/1.3681140.
The time- and length-scale accessible to molecular dynamics simulations of biomolecular systems using atomic-level (AL) models is most limited by the calculation of the solvent-solvent interactions, which comprise the majority of the interactions and yet are seldom of specific interest. Coarse-graining (CG), in which multiple solvent molecules are subsumed into a single bead, provides a means of overcoming this limitation without resorting to implicit solvation models, which basically misrepresent the hydrophobic effect. Most existing CG models, however, do not explicitly include electrostatic interactions, and thus fail to reproduce important properties of the solvent such as dielectric screening. Moreover, CG models for one type of solvent molecule are seldom compatible with those for other solvents. Here, we develop polarizable CG models for the solvents dimethyl sulfoxide, chloroform, and methanol that are compatible with an existing CG model for water. The inclusion of polarizability greatly improves the reproduction of thermodynamic data measured experimentally and calculated from AL simulations for both the pure liquids and binary mixtures.
使用原子级（AL）模型对生物分子系统进行分子动力学模拟的时间和长度尺度受到限制，主要是因为溶剂-溶剂相互作用的计算，溶剂-溶剂相互作用占相互作用的大部分，但很少有具体的兴趣。粗粒化（CG）将多个溶剂分子归入一个珠子中，提供了一种克服这种限制的方法，而无需诉诸基本上无法正确表示疏水作用的隐式溶剂化模型。然而，大多数现有的 CG 模型并不明确包括静电相互作用，因此无法再现溶剂的重要性质，例如介电屏蔽。此外，一种溶剂分子的 CG 模型很少与其他溶剂的 CG 模型兼容。在这里，我们为二甲亚砜、氯仿和甲醇这三种溶剂开发了可极化的 CG 模型，它们与现有的水 CG 模型兼容。可极化性的包含极大地提高了对纯液体和二元混合物的实验测量和从 AL 模拟计算的热力学数据的再现性。
