Estimation of absolute solvent and solvation shell entropies via permutation reduction.
Estimation of absolute solvent and solvation shell entropies via permutation reduction.
复制标题
通过置换约简估计绝对溶剂和溶剂化壳熵。
DOI：
10.1063/1.2400220
复制
发表时间：
2007
期刊：
影响因子：
--
通讯作者：
H. Grubmüller
中科院分区：
文献类型：
--
作者：
F. Reinhard;H. Grubmüller
尽管溶剂化壳层的熵对蛋白质或DNA等溶剂化大分子的自由能有显着贡献，并且主要包含在分子动力学模拟中，但溶剂化壳层的熵无法直接应用已建立的熵估计方法来获得。复杂性是双重的。首先，此类系统的构型空间密度过于复杂，无法实现足够精确的拟合。其次，与内部大分子动力学相反，溶剂分子扩散运动探索的构型空间体积太大，无法通过当前的模拟技术进行详尽的采样。在这里，我们开发了一种方法来克服第二个问题并显着缓解第一个问题。我们建议通过变换轨迹来利用溶剂的排列对称性，使已建立的估计方法适用，例如准调和近似或主成分分析。我们的排列减少方法涉及一个组合问题，该问题通过其与线性分配问题的等价性来解决，线性分配问题存在 O(N3) 方法。通过稠密伦纳德-琼斯气体的测试模拟，获得了增强的收敛性和改进的熵估计。此外，我们的方法使扩散系统可以改进拟合函数。
Despite its prominent contribution to the free energy of solvated macromolecules such as proteins or DNA, and although principally contained within molecular dynamics simulations, the entropy of the solvation shell is inaccessible to straightforward application of established entropy estimation methods. The complication is twofold. First, the configurational space density of such systems is too complex for a sufficiently accurate fit. Second, and in contrast to the internal macromolecular dynamics, the configurational space volume explored by the diffusive motion of the solvent molecules is too large to be exhaustively sampled by current simulation techniques. Here, we develop a method to overcome the second problem and to significantly alleviate the first one. We propose to exploit the permutation symmetry of the solvent by transforming the trajectory in a way that renders established estimation methods applicable, such as the quasiharmonic approximation or principal component analysis. Our permutation-reduced approach involves a combinatorial problem, which is solved through its equivalence with the linear assignment problem, for which O(N3) methods exist. From test simulations of dense Lennard-Jones gases, enhanced convergence and improved entropy estimates are obtained. Moreover, our approach renders diffusive systems accessible to improved fit functions.