Supercomputing Studies of Biomembranes

Abstract

Although computer simulation of biological molecules has seen widespread growth and is widely accepted as an important biochemical tool, it is hampered by lim ited computing resources. Biomolecular systems, by necessity, contain a large number of interaction sites. In many cases, these sites interact over quite large dis tances. Further, the time scales of biological interest are long, which requires that simulations of dynamical properties at the atomic level must be lengthy to ade quately probe these motions. We address these issues through discussions of atomic-level molecular dynam ics simulations of biological lipid bilayer membranes, which are key constructs in biochemistry. These simu lations reproduce many experimental observables and provide a degree of resolution currently unavailable experimentally. The lengths of these simulations, the longest of which was 2 nanoseconds, were sufficient to effectively sample many of the motions governing the behavior of biomembranes. Examples are given showing the importance of long-range interactions. The number of interaction sites required by these sim ulations is discussed, particularly the need for explicit representation of solvent molecules.