The Research On Energy Model Based Otimization Method Of Protein Conformational Space

Three dimensional structure of protein is the key to understand and transform the biological and cytological functions.However,the analytical speed of the three dimensional structure of protein is far less than that of the new sequence.Therefore,it is a crucial problem to predict the three-dimensional structure of proteins from amino acid sequences in bioinformatics.According to the research status at home and abroad,the sampling efficiency can be improved by using the global search abilityof the differential evolution algorithm and combining the fragment assembly with the replica exchange.In addition,Monte Carlo method also is employed for population improvement.Then the decoy conformation distance model is designed according to the distance profile.It is been introduced into the selection operation of the algorithm based on a normal distribution probability.The evolve direction of population can be guide to detect the area with low energy and reasonable structure.The main work and the research results are as follows:(1)The background and significance on the prediction of protein 3D structure is firstly summerized.Then the basic concepts and development process of the modern evolutionary algorithms and energy models are introduced.Specially,the research and application of the evolutionary algorithm in protein structure prediction is described in detail.(2)To solve the problem of insufficient sampling ability in protein structure prediction method,a replica exchange strategy is introduced based on differential evolution algorithm framework.A conformation population is put into every replica layer.Then the differential evolution algorithm is adopted to update the population in each layer.Finally,the updated populations are enhanced by Monte Carlo method.As a consequence,the global optimal conformation and a series of metastable conformations are generated.The conformational space can be globally searched,which takes advantage of the strong global searching ability of differential evolution algorithm.The good local searching performance of Monte Carlo is also employed to sample the local minimum area adequately.During the search process,Replica exchange strategy ensures the diversity of population in replica layers,and the capacity of the algorithm to jump out of local minimum is enhanced as well.Therefore,the searching ability is further heightened.Test results of 10 target proteins show that the proposed method can generated high-resolution near-native protein conformations by searching the conformational space rapidly and effectively.(3)The ab-initio protein structure prediction excessively depending on the inaccurate energy model in the optimization process of high-dimensional conformational space,which is a key element that restricts the accuracy of prediction.A searching method using knowledge-based distance profile in ab-initio protein structure prediction(DPKM)is proposed based on the framework of differential evolution algorithm.Non-redundant template library is constructed based on the Protein Data Base,and then the residue-residue distance profile can be constructed through the fragment library obtained by the gapless threading.Then decoy conformation distance model is designed according to the distance profile,and been introduced into the selection process of the algorithm by utilizing a normal distribution probability.The fragment assembly also is used to improve the efficiency and accuracy of the prediction.Take advantage of the global searching ability of DE algorithm,the evolve direction of population can be lead to the area with low energy and reasonable structure under the double restriction of the energy and distance.Test results of 20 target proteins show that the proposed method can obtain near-native protein conformations by sampling the conformational space rapidly and effectively.