Efficient sampling of high-dimensional free energy landscapes using adaptive reinforced dynamics

Abstract

Enhanced sampling methods such as metadynamics and umbrella sampling have become essential tools for exploring the configuration space of molecules and materials. At the same time, they have long faced the following dilemma: Since they are only effective with a small number of collective variables (CVs), choosing a proper set of CVs becomes critical for their accuracy. In this work, we show that with some technical improvements, reinforced dynamics (RiD) can be used to efficiently explore the configuration space and free energy landscapes with a large number of CVs, thereby alleviating the difficulty associated with choosing two or three CVs. We illustrate this by studying various representative and challenging examples. As the first example, we construct the 9-dimensional free energy landscape of the peptoid trimer which has energy barriers of more than 8 kcal/mol. We then study the folding of the protein chignolin using 18 CVs. With a small computational budget, we are able to observe 51 transitions between the folded and unfolded states. Finally, we propose a new protein structure refinement protocol based on RiD. This new protocol allows us to efficiently employ more than 100 CVs for exploring the landscape of protein structures and it gives rise to an overall improvement of 14.6 units over the initial GDT-HA score.

Insights into glyphosate removal efficiency using a new 2D nanomaterial

“…The WT-MtD simulation is implemented with GROMACS 2019.2 and its plugin PLUMED (version 2.5.2) for 100 ns. 33,40 The FES is explored as a function of the distance between the center of mass (COM) of GLY and SNS surface. More details about the WT-MtD simulation are provided and analyzed in the "Metadynamics results" section.…”