GA-guided mD-VcMD: A genetic-algorithm-based method for multi-dimensional virtual-system coupled molecular dynamics

TL;DR

This paper introduces a novel genetic-algorithm-guided extension to the multi-dimensional virtual-system coupled molecular dynamics (mD-VcMD) method, enabling efficient conformational sampling and thermodynamic analysis of biomolecular systems without extensive fine-tuning.

Contribution

The paper presents the GA-based mD-VcMD, an innovative extension that improves sampling efficiency and analysis flexibility in biomolecular simulations.

Findings

The method allows use of all simulation snapshots for analysis.

It eliminates the need for fine-tuning weight functions.

It enables selective sampling of poorly-sampled regions.

Abstract

We previously introduced a conformational sampling method, a multi-dimensional virtual-system coupled molecular dynamics (mD-VcMD), to enhance conformational sampling of a biomolecular system by computer simulations. Here, we present a new sampling method, subzone-based mD-VcMD, as an extension of mD-VcMD. Then, we further extend the subzone-based method using genetic algorithm (GA), and named it the GA-based mD-VcMD. Because the conformational space of the biomolecular system is vast, a single simulation cannot sample the conformational space throughout. Then, iterative simulations are performed to increase the sampled region gradually. The new methods have the following advantages: (1) The methods are free from a production run: I.e., all snapshots from all iterations can be used for analyses. (2) The methods are free from fine tuning of a weight function (probability distribution function or potential of mean force). (3) A simple procedure is available to assign a thermodynamic weight to snapshots sampled in spite that the weight function is not used to proceed the iterative simulations. Thus, a canonical ensemble (i.e., a thermally equilibrated ensemble) is generated from the resultant snapshots. (4) If a poorly-sampled region emerges in sampling, selective sampling can be performed focusing on the poorly-sampled region without breaking the proportion of the canonical ensemble. A free-energy landscape of the biomolecular system is obtainable from the canonical ensemble.