Research indexJournal article
Parameter Optimization\nMethod in Multidimensional\nUmbrella Sampling
(5898206), (1777582)
Figshare5 Aug 2024Open access
NobleIDark:/48914/w1/20260525/19AC80B3DOI10.1021/acs.jctc.4c00282.s001
Parameter Optimization\nMethod in Multidimensional\nUmbrella Sampling
(5898206), (1777582) · Figshare · 2024
NobleIDark:/48914/w1/20260525/19AC80B3DOI10.1021/acs.jctc.4c00282.s001
Abstract
Umbrella sampling (US) is an effective method for calculating\nfree-energy\nlandscapes (FELs). However, the complexity of controlling the sampling\npositions complicates multidimensional FEL calculations. In this study,\nwe proposed a method for controlling sampling by optimizing the US\nparameters. This method comprises the introduction of a target point\nand the optimization of the parameters to sample a window around this\npoint. We approximated each window to normal distributions using an\numbrella integration method and calculated the divergences between\nthe window distributions and the state distributed at the target position\nby a variationally enhanced sampling method. Thus, the minimization\nof the divergence facilitated sampling around the target point, after\nwhich the parameters could be optimized on the fly while performing\nequilibration simulation. In practice, our method employs bias potentials\nwith off-diagonal terms, ensuring a more efficient calculation of\nmultidimensional FEL. Additionally, we developed an algorithm for\ndetermining the target point for automated FEL search; the algorithm\nsamples in a specified direction while controlling the overlap of\ndistributions. We performed three different FEL calculations as examples:\n(1) the calculation of the permeation of a water molecule through\na lipid bilayer (one-dimensional FEL), (2) the calculation of the\ninternal structural changes in alanine dipeptide in water (two-dimensional\nFEL), and (3) the calculation of the internal structural changes from\na β-strand structure to an α-helix structure in alanine\ndecapeptide (Ala10, 16-dimensional FEL). These results confirmed that\nour method could control the number of US windows and calculate the\nhigh-dimensional FELs that could not be evaluated by the conventional\nUS method.