Department of Chemistry, James Franck Institute, and Institute for Biophysical Dynamics , The University of Chicago , Chicago , Illinois 60637 , United States.
J Chem Theory Comput. 2018 Dec 11;14(12):6159-6174. doi: 10.1021/acs.jctc.8b00812. Epub 2018 Nov 16.
Coarse-graining (CG) methodologies have been widely used to extend the time and length scales of computer simulations by averaging over the atomistic details beneath the resolution of the CG models. Despite the efficiency of CG models, important configurational information during a given process may be lost at the CG resolution. One example of this is the topology of the hydrogen bonding network in the liquid state. When the functional group that participates in hydrogen bonding (e.g., -OH in methanol) is coarse-grained into one CG site, the effective interactions of the resultant CG model are usually derived from an averaged overall trajectory and, thus, do not take into account the hydrogen bonding interactions and topologies that are present at the all-atom resolution. In order to overcome this challenge, the present study develops new ultra-coarse-grained (UCG) models that include internal states within the CG sites that participate in hydrogen bonding, where each state represents a specific configuration such as the hydrogen bonding donor or acceptor. Internal states of the UCG beads are modeled to remain in quasi-equilibrium, and the degree of mixing is controlled by utilizing the effective local density of the UCG sites. In particular, we consider two groups of UCG models with different types of hydrogen bonding motifs: chain-like and ring-like. Using five different liquid systems that contain the same fundamental functional groups as biomolecules, we demonstrate the ability of the UCG models to reproduce the structural properties that originate from the configurations beneath the resolution of the UCG model. This proposed approach can also be applied to other liquids with such specific and directional interactions, or even to complex biomolecular systems in which hydrogen bonding is critical.
粗粒化 (CG) 方法被广泛用于通过在 CG 模型的分辨率以下的原子细节上进行平均来扩展计算机模拟的时间和长度尺度。尽管 CG 模型具有效率,但在给定过程中,重要的构型信息可能会在 CG 分辨率下丢失。这种情况的一个例子是液体状态下氢键网络的拓扑结构。当参与氢键的官能团(例如甲醇中的 -OH)被粗粒化为一个 CG 位点时,所得 CG 模型的有效相互作用通常是从平均整体轨迹中得出的,因此不会考虑氢键相互作用和在全原子分辨率下存在的拓扑结构。为了克服这一挑战,本研究开发了新的超粗粒化 (UCG) 模型,这些模型在参与氢键的 CG 位点中包含内部状态,其中每个状态代表特定的构型,如氢键供体或受体。UCG 珠的内部状态被建模为处于准平衡状态,并且通过利用 UCG 位点的有效局部密度来控制混合程度。特别是,我们考虑了具有不同氢键模体类型的两组 UCG 模型:链状和环状。使用包含与生物分子相同基本官能团的五个不同液体系统,我们展示了 UCG 模型重现源自 UCG 模型分辨率以下构型的结构性质的能力。这种方法还可以应用于具有这种特定和定向相互作用的其他液体,甚至可以应用于对氢键至关重要的复杂生物分子系统。