Highlights of our Work
Olga Svinarski and VMD
Water is ubiquitous; living cells and the cells' biomolecules are greatly influenced by water. Accounting for water in computational modeling, a key access route to imaging cell processes, comes at a stiff price: up to 90% of the simulation volume needs to be a water bath embedding simulated biomolecular machinery. The sheer number of water molecules arising here, sometimes millions, and the strong hydrodynamic drag resisting biomolecular dynamics, slows down simulations tremendously. Computational biologists had invented a means to replace water by a kind of continuous ether, so-called implicit solvent, that accounts for key electrostatic features of water around biomolecules, but does not slow down simulations. Computational hinderance through explicit water is getting worse with increasing size of biomolecular machines, but so far the implicit solvent description could not be effectively employed to model large systems using supercomputers. In a recent study, the situation has been repaired and the use of an implicit solvent in molecular dynamics simulations of systems as large as the whole ribosome (containing 300,000 ribosome atoms and eliminating 2,700,000 water atoms) have been documented. In a further study, the implicit solvent description has been extended to account also for hydrophobic interactions at biomolecular surfaces; the extension took advantage of delegating calculations both to GPU and CPU as the latter is more suitable for the evaluation of surface interactions. More information regarding NAMD's parallel computer-ready implicit solvent can be obtained from the NAMD website.