| The Protein folding structure prediction is one of the theoretically fundamental issues in Bioinformatics. It is significant to solve this problem to related disciplines such as being cognition, pharmacy and illness treatment. But it is unavailable to detect the real structures of protein applying today's apparatus and method. So scholars suppose the idea"sequence decide structure and structure decide function", abstract and simplify the protein structure, change the issue into mathematics question, and resolve it through modern computer science and technology.The research to simplified model is of great importance. At present, there are mainly HP model, AB model and continual model. Aim at these leading models, the scientists on computer science and biology carry out a great deal researches. No matter what kind of the model is, the resolvent will be ascribed to NP-Hard problem on computer theory science at last. The relevant methods are Monte-Karlo Method, Simulate Anneal Arithmetic etc.The research on the 3D-Structure of Protein is important. Firstly, according to the research results given by biologist recently and the abstract of the problem of Protein, a related arithmetic model is found through quasiphysical ways which change a limited problem into an unlimited problem. Then solved the 3D arithmetic model, chiefly discussed three aspects: the selection about initialization of Protein structure, the calculation of the best structure on local position, the jumping strategy. At last, a method of gradient descent by advanced step to get the local minimum is formulated and some good jumping strategy to get the integral minimum are also given out.On the basis of ideas of 3D-structure arithmetic model and resolvents, a concrete quasiphysical and anthropopathic algorithm is formulated. Through calculating by computer and comparing with the former results, it is proved that the quasiphysical and anthropopathic algorithm is effective to the research for Protein 3D-Structure.
|
PDF Full Text Request