Computational modeling of protein interactions at multiple lengthscales

摘要: We developed theories and algorithms for two coarse-grained implicit solvent models that can be deployed within a multiscale framework to enable computational studies of large-scale protein-protein associations. The first model is residue level alpha-carbon bead intended simulating proteins at close range during formation encounter complexes. This introduces novel forcefield term directional backbone hydrogen bond semi-explicitly, as well fourth flavor in its sequence-dependence better represent the spectrum residue-residue attractive interactions. showed introduction orientation-dependent bonding resulted more stable realistic alpha helices beta sheets. In addition, addition reduces energetic frustrations competition from misfolded states. overall increased folding cooperativity, greater structural faithfulness experimentally solved structures. efficiency has also permitted us develop molecular Alzheimer's A-beta 1-40 fibril study nucleation elongation, providing good proof-of-concept laying foundation applications other assembly processes. second protein diffusional search. It treats rigid bodies interacting solely through long-range electrostatics. described theory implementation method, Poisson-Boltzmann Semi-Analytical Method (PB-SAM), electrostatic interactions by efficiently solving linearized equation (PBE). method combines advantages analytical boundary element methods representing macromolecular surface realistically collection overlapping spheres, which polarization charges then iteratively using multipole method. Unlike finite difference solvers, PB-SAM not constrained spatially box size, making it suitable dynamics. this realizes accuracy reduced cost relative either or PBE solvers. derived expressions force torque account mutual both zero order derivative charges, incorporated complete into Brownian dynamics simulation algorithm. demonstrated time dynamic propagation multiple particles with accurate accounting effects successive timesteps, system monomers brome mosaic virus (PDB code: 1YC6). While systems hitherto inaccessible spatial dimensions, we further reduce computation parallelization, faster linear algebra operations, optimizing convergence criteria cutoffs, approximating models. Finally, discussed strategies connect above studies. employed successively nested variant Northrup-Allison-McCammon formalism compute bi-molecular kinetics rates. kinetic parameters turn inputs chemical master equations stochastic simulations. Such modeling used determine rates association, help investigate how changing physical alter association rates, consequently control sequences association.