Paper title: Scalable Molecular Dynamics with NAMD
Journal: Journal of Computational Chemistry (2005) (Vol. 26, No. 16)
Author(s): Phillips, J.C.[1], Braun, R.[1], Wang, W.[1], Gumbart, J.[1], Tajkhorshid, E.[1], Villa, E.[1], Chipot, C.[2], Skeel, R.D.[3], Kale, L.[3] and Schulten, K.[1]
[1] Beckman Institute, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801
[2] UMR CNRS/UHP 7565, Universite Henri Poincare, 54506 Vandoeuvre-les-Nancy, Cedex, France
[3] Department of Computer Science and Beckman Institute, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801
Summary:
The paper may be broken down into two major parts. First, it explores the theories, concepts and algorithms involved in molecular dynamics (MD). Then, it gives details about the applicability of the parallel program, NAMD, in MD simulations by looking deeper into its design and effective usage in parallel computing machines and by providing sample simulations.
NAnoscale Molecular Dynamics, or more popularly known as NAMD, is a parallel molecular dynamics code designed for high-performance simulation of large biomolecular systems. It can scale to hundreds of processors on high-end parallel platforms and tens of processors on commodity clusters with gigabit ethernet. It is used alongside with VMD (Visual Molecular Dynamics), a molecular visualization program that displays, animates, and analyzes biomolecular systems using 3D graphics and built-in scripting. It is distributed free of charge and is well documented for ease of use.
Molecular dynamics simulation (MDS) assumes that molecules move according to the Newtonian equation of motions. This gives a relationship between the mass of the atom, its position and the total potential energy based on all atomic positions. It is also assumed that each atom experiences a force specified by a model force field (through potential energy) that accounts for the interaction of that atom with the rest of the system. For non-bonded or long-range interactions, full electrostatic computations are possible with NAMD using the smooth particle-mesh Ewald summation (SPME) method. All equations are then numerically integrated by NAMD using the velocity-Verlet method to obtain the position and velocity for the next time step using the current ones.
The paper also discussed about other approaches that can be done with MD. Statistical mechanics methods may be used to simulate systems where temperature or pressure is controlled (NVT and NPT ensembles). A specific external force may be applied to molecules through Steered MD (SMD) and Interactive MD (IMD). Using NAMD and VMD, both SMD and IMD are able to probe the mechanical properties of molecules or to accelerate processes that are very slow to model. Furthermore, aside from simulating the motions of atoms, MD may also be used to calculate thermodynamic quantities like free energy.
The design and implementation of NAMD as well as its scalability to be used in parallel computers were explained. As a demonstration of the capabilities of the software, the authors cited three examples of how to apply NAMD. First, they applied it with a small protein system, ubiquitin, using SMD. The modeling task for this case is to explain atomic force microscopy experiments that measured forces and extensions that arise when ubiquitin is stretched with pN forces. Afterwards, they tried simulating larger systems using aquaporin (membrane protein) and a protein-DNA complex.
Contribution and application:
Molecular dynamics simulations have long been widely used in the field of life sciences to study proteins and other cell structures. In the same manner, MDS can also be used by ESEL to shed some light on the complicated fouling mechanism and movement of particles during SWRO desalination process.
By: Hannah Ebro
hannah@gist.ac.kr