Applications of Parallel Computing to Biological Problems
Boris Ostrovsky, Mark A. Smith, Yaneer Bar-Yam · Annual Review of Biophysics and Biomolecular Structure · 1995
Parallel computers should provide the greatest processing power and memory for scientific simulations in the coming decades. This review discusses general strategies and specific algorithms for the use of various parallel architectures in simulations of biological and artificial polymers. General strategies include space partitioning (domain decomposition cell methods) and distributed independent simulations. Specific algorithms include cellular automata for efficient abstract polymer simulation. One algorithm, the two-space algorithm, is particularly efficient both for parallel and serial computation. Three applications, 2D melts, gel electrophoresis, and polymer collapse, are described. Simulations of high-density melts in 2D show that contrary to expectations, polymers do not completely segregate at the highest densities; instead, polymer interpenetration is significant. Preliminary simulations of gel electrophoresis show its behavior in the diffusive regimen and demonstrate the use of Cellular Automaton Machines (CAMs). Polymer collapse is studied in the regime of large departures from good solvent conditions. In this regime, kinetics plays a significant role. Collapse is dominated (nucleated) by migration of the chain ends.