New theoretical techniques for studying physical properties of proteins have been developed and characterized. These are done in conjunction with software development, and involve the systematic testing and evaluation of new ideas. Specific projects include: -Development of Langevin piston methods for NPT simulation of periodic systems and for stochastic boundary molecular dynamics (MD) simulations -Development of quantum mechanical potentials and appropriate algorithms for use in MD -Development of an optimized protocol for the preparation of low-temperature states -Development of flexible MD techniques that remove high frequency degrees of freedom -Development of the REPLICA/PATH method for determining reaction paths in complex systems using simulated annealing -Free energy perturbation simulations in solution, examining the effect of restraints -Adaptation of a truncated Newton minimizer for CHARMM and biomolecular applications -Symplectic methods for molecular dynamics simulations -Development of constant pressure stochastic boundary methods -Rational drug design: shape descriptor facility for CHARMM -Sidechain placement, insertions, and deletions iin prediction by homology Although many of the parameter sets and models that are generally available are of the quality required for accurate simulation of macromolecular systems, there remains the need to weigh the relative merits of these sets for the specific types of systems studied in the MGS. There are also numerous details such as the treatment of water and long-range electrostatics. Ongoing projects include: -Evaluation of parameter sets -Approximation of long-range interactions in macromolecular simulation variants of the Ewald sum method -New methods for long-range truncation of the energy potential -Evaluation and comparison of implicit and explicit water models for simulations examining proteins hydration -Development and use of a polarized and flexible water model -Three-dimensional structure determination of proteins from a simiplified topological description -Conversion of physical models into three-dimensional coordinates for computer analysis and simulation
