Generation of Pathway Signatures by Combining Normal Modes with Weighted Ensemble Simulations

Abstract

Molecular dynamics simulations provide a "computational microscope" by which molecular phenomena can be studied at atomic resolution. However, such simulations are often expensive, usually due to a combination of system size and timescale. Various enhanced sampling methods have been proposed to overcome these challenges. Despite their effectiveness, many suffer from artifacts from energetic biases guiding the simulations, or lack of effective progress coordinates. Proteins normal modes uniquely defined by their 3D fold capture their intrinsic dynamics and could provide unbiased guidance, but how to combine these modes with molecular dynamics to generate continuous, energetically unbiased pathways has been challenging. In this study, we demonstrate that conformations generated along from normal modes using adaptive anisotropic network model provide a physical, intuitive, and generalizable progress coordinate for weighted ensemble simulations, providing a boost in efficiency and a means to generate pathways for any protein system without prior knowledge.