Bereau Tristan, Kremer Kurt
Max Planck Institute for Polymer Research , Ackermannweg 10, 55128 Mainz, Germany.
J Chem Theory Comput. 2015 Jun 9;11(6):2783-91. doi: 10.1021/acs.jctc.5b00056. Epub 2015 May 12.
The systematic exploration of chemical compound space holds many promises toward structure-function relationships and material design. In the context of computer simulations, progress is hampered by both the sheer number of compounds and the efforts associated with parametrizing a force field for every new molecule. A coarse-grained (CG) representation provides not only a reduced phase space but also a smaller number of compounds, due to the redundancy of CG representations mapping to the same structure. Though many CG models require the explicit force-field parametrization of a molecule with all others, others assume transferability by means of mixing rules, such as the Martini force field. To alleviate the burden associated with tedious parametrizations for each new compound, the present work aims at automating the mapping and parametrization of common small organic molecules for Martini. We test the method by analyzing the water/octanol partitioning of more than 650 neutral molecules, the hydration free energy of 354 others, and the free energies of hydration and solvation in octanol of another 69 compounds. Last, we compare with all-atom simulations the thermodynamics of insertion of four individual solute molecules in a phospholipid membrane. The protocol demonstrates the feasibility of an automated parametrization scheme for Martini and provides prospects for high-throughput simulation methodologies.
对化合物空间的系统探索在揭示结构-功能关系和材料设计方面有着诸多前景。在计算机模拟的背景下,化合物数量之多以及为每个新分子参数化力场所需的工作量阻碍了研究进展。粗粒度(CG)表示不仅提供了一个简化的相空间,而且由于CG表示映射到相同结构时存在冗余,化合物数量也较少。尽管许多CG模型要求对每个分子与其他所有分子进行显式的力场参数化,但其他一些模型则通过混合规则(如马天尼力场)假设具有可转移性。为了减轻为每个新化合物进行繁琐参数化带来的负担,本工作旨在实现马天尼通用小有机分子的映射和参数化自动化。我们通过分析650多个中性分子的水/辛醇分配、另外354个分子的水合自由能以及另外69种化合物在辛醇中的水合和溶剂化自由能来测试该方法。最后,我们将四个单个溶质分子插入磷脂膜的热力学与全原子模拟进行比较。该方案证明了马天尼自动化参数化方案的可行性,并为高通量模拟方法提供了前景。