Grid-based nonequilibrium multiple-time scale molecular dynamics/Brownian dynamics simulations of ligand-receptor interactions in structured protein systems

Abstract

In the hybrid Molecular Dynamics (MD)/Brownian Dynamics (BD) algorithm for simulating the long-time, nonequilibrium dynamics of receptor-ligand interactions, the evaluation of the force autocorrelation function can be computationally costly but fortunately is highly amenable to multimode processing methods. In this paper, taking advantage of the computational grid's large-scale computational resources and the nice characteristics of grid-based Monte Carlo applications, we developed a grid-based receptor-ligand interactions simulation application using the MD/BD algorithm. We expect to provide high-performance and trustworthy computing for analyzing long-time dynamics of proteins and protein-protein interaction to predict and understand cell signaling processes and small molecule drug efficacies. Our preliminary results showed that our grid-based application could provide a faster and more accurate computation for the force autocorrelation function in our MD/BD simulation than previous parallel implementations.