Okur Asim, Roe Daniel R, Cui Guanglei, Hornak Viktor, Simmerling Carlos
Department of Chemistry and Center for Structural Biology, Stony Brook University, Stony Brook, New York 11794, and Computational Science Center, Brookhaven National Laboratory, Upton, New York 11973.
J Chem Theory Comput. 2007 Mar;3(2):557-68. doi: 10.1021/ct600263e.
Parallel tempering or replica-exchange molecular dynamics (REMD) significantly increases the efficiency of conformational sampling for complex molecular systems. However, obtaining converged data with REMD remains challenging, especially for large systems with complex topologies. We propose a new variant to REMD where the replicas are also permitted to exchange with an ensemble of structures that have been generated in advance using high-temperature MD simulations, similar in spirit to J-walking methods. We tested this approach on two non-trivial model systems, a β-hairpin and a 3-stranded β-sheet and compared the results to those obtained from very long (>100 ns) standard REMD simulations. The resulting ensembles were indistinguishable, including relative populations of different conformations on the unfolded state. The use of the reservoir is shown to significantly reduce the time required for convergence.
并行回火或副本交换分子动力学（REMD）显著提高了复杂分子系统构象采样的效率。然而，使用REMD获得收敛数据仍然具有挑战性，特别是对于具有复杂拓扑结构的大型系统。我们提出了一种REMD的新变体，其中副本也被允许与预先使用高温分子动力学模拟生成的一组结构进行交换，这在本质上类似于J-walking方法。我们在两个非平凡的模型系统上测试了这种方法，一个β-发夹和一个三链β-折叠，并将结果与从非常长（>100纳秒）的标准REMD模拟中获得的结果进行了比较。得到的系综无法区分，包括未折叠状态下不同构象的相对丰度。结果表明，使用储库可以显著减少收敛所需的时间。
Zotero 插件