(272d) A Recent Advance in Modeling Dynamic Properties With Coarse-Grained Molecular Simulation

Authors: 
Markutsya, S. - Presenter, Iowa State University
Lamm, M. H., Iowa State University



Coarse-grained models that are rigorously derived from all-atom molecular simulation trajectories can be successfully implemented in molecular dynamics simulation to reproduce the structure and thermodynamics of the all-atom reference system. Dynamic properties from these coarse-grained molecular simulations are often unreliable because the reduction in degrees of freedom eliminates much of the friction between coarse-grained beads. This results in faster dynamics for the coarse-grained simulation. This is an advantage when it comes to efficient sampling of phase space but problematic if accurate computation of one or more time-dependent properties is a desired outcome from the coarse-grained simulation. In this work, a new approach is developed for deriving coarse-grained intermolecular forces. This approach retains the frictional contribution that is often discarded by conventional coarse-graining methods. This method is based on the well-known Langevin equation formalism. Compared to previous implementations of the Langevin equation for coarse-grained dynamics, in this method, the friction coefficients are computed directly during the derivation of the CG force field, using routine post-processing calculations. The procedure is tested for water and an aqueous glucose solution, and the results from the new implementation for coarse-grained molecular dynamics simulation show remarkable agreement with the dynamics obtained from reference all-atom simulations. The agreement between the structural properties observed in the coarse-grained and all-atom simulations is also preserved. We discuss how this approach may be applied broadly to any existing coarse-graining method where the coarse-grained models are rigorously derived from all-atom reference systems.

This research is sponsored by the U.S Department of Energy (USDOE) Scientific Discovery through Advanced Computing (SciDAC) program through USDOE's Office of Advanced Scientific Computing Research (ASCR) and Biological and Environmental Research (BER), and performed at the Ames Laboratory, FWP AL-08-330-039.