Routine Microsecond Simulate The Folding Process Of Fast-folding Proteins

In recent years in the molecular dynamics simulation research, protein folding simulation studies have made great progress by applied to combines GPU, new force field (ffl2SB), and new implicit solvent model (GB-Neck2). The special hardware of GPU have traditionally been used in accelerated graphics operation, it is different from the general purpose calculative ability in particular case. NVidia CUDA programming tools that GPU can much better than the CPU computing power, CUDA stands for Compute Unified Device Architecture and is a new hardware architecture for issuing and managing computations on the GPU as a data-parallel computing device and programming model. A number of important algorithms have been used in GPU, one or two orders of magnitude higher than CPU computing performance. The biological molecules need large amount of calculation in the molecular dynamics simulation, the GPU is the best choice for the calculation of molecular dynamics. The new field (ff12SB) has been able to reasonable accurately describe the interaction between atoms. The implicit solvent framework is based on replacing the real water environment consisting of discrete molecules by infinite continuum solvent, the solvation free energy describe by the solvent effect. Although based on implicit solvent framework make some basic approximation to the realistic bio-molecular simulation, but now the development of implicit solvent water model can well describe the solvent effect, such as the generalized Born (GB) model, the physical foundations of the model and its derivation from the underlying Poisson-Boltamann (PB) model. This approximation model has several advantages compared to the explicit solvent framework effect, especially in molecular dynamics simulations. These include:lower direct computational costs for many molecular systems; enhanced sampling of conformational space; effective ways to estimate free energies. Availability of all of these advantages on practical computational models such model that has become particularly popular.In previous studies, the α-helix folding simulation more easily than β-sheet, due to the α-helix can make folding in nanoseconds, β-sheet need time is longer and it folded microsecond range. In this thesis, with the new design protocol for α-helix, β-sheet folding simulations, and has folded conformation.This thesis is organized as follows. The basic content of the basic process of protein folding, folded features and model is introduced section one. The basic theories of molecular dynamics simulation theory, the potential energy function of biological molecules, solvent model and conformational analysis are presented in section two. A fast protocol to simulate the folding process of fast-folding proteins is presented in section three. The conclusions of the paper are gathered in section four.Among of the three sections is our main work during the period of postgraduate. We performed molecular dynamics (Langevin) folding simulation for four proteins, CLN025(2ZEI), MHA6(2I9M), Trp-Cage (1L2Y) and Villin (3TRW), using AMBER ff12SB force field and Generalized Born (GB-Neck2) implicit solvent model on GTX670GPU with AMBER12package. The results suggest that our new method can also make it complete folded in microsecond time range, the compare results of between simulation structure and native structure computing RMSD not exceeding2A: The folding time is5.94μs and the RMSD compared to its crystal structure is0.897A for CLN025in our molecular dynamics simulation,0.191μs and1.142A for MHA6;4.23μs and1.37A for Villin,2.49μs and0.63A for Trp-Cage, respectively. The results suggest that the simulation of microsecond protein folding process can be reached on a desktop workstation with GPU.