Efficient Conformational Sampling For Peptides And Proteins

The computer simulation has been widely used into the study of biomacromoleculessuch as proteins, however, the poor sampling of conventional molecular dynamicssimulation of peptides and proteins is usually considered to be due to numerous localminima in the free energy surface, and during the sampling, it will always stay in theselocal minima state, yet can’t go to the other states around quickly.Our work begins with the essential composition of proteins, and reviewing thetwenty different amino acids making the proteins, then we get back to study the physicalproperties, including free energy. The next, we study the proteins using the moleculardynamics simulation methods, and then we proposed an efficient double couplingmethod, At the end, we apply this method to one specific polypeptide to verify ourmethod.At the same time, this paper will show that it is the unfavorable distribution ofkinetic energy along special degrees of freedom which should be responsible for thispoor sampling. This is due to the decaying form of the kinetic energy distribution alongeach degree of freedom. To overcome these problems, we modified the kinetic energydistribution by further coupling the environment thermal bath onto the critical orderparameters separately and changed the distribution of the kinetic energy along theseorder parameters to a Gaussian shape.In our work, we apply this method to ALA dipeptide model, and we find that thismodified molecular dynamics greatly enhances the sampling behavior in theconformational space and constructs a complete free energy surface along its backbonetorsion angles, while conventional molecular dynamics fails to do so.