Research Of Secondary Structure And Residue-Residue Contact Based Protein Structure Prediction Method

Proteins are the building blocks of life,and the function of a protein depends on its tertiary structure.Protein structure is very important in protein research,only with the protein structures known,it is possible to design drugs which can interact with the protein surface and further change its bad behavior,to locate mutations in the protein structure that lead to disease.Generally,in the field of protein structure prediction,there are two bottleneck problems,such as inaccuracy of the model and huge conformational space.To design efficient conformational space optimization method based on prior knowledge is an important problem in the field of protein structure prediction.Based on the framework of evolution algorithm,an improved differential evolution protein structure prediction method is proposed with the use of secondary structure and residue-residue contact information,namely,SCDE.The main work as follows:1.Based on the review of research at home and abroad,the key technologies,such as protein structure,residue-residue contact map,fragment assembly,and energy scoring function are introduced in detail.In addition,the Rosetta platform and model accuracy measurement are also introduced.2.Based on the predicted secondary structure,the similarity model is built,selection strategy of evolutionary algorithm is designed,thus the orientation searching of secondary structure in huge conformational space is implemented.With the use of predicted residue-residue contact,a protein structure model is constructed;The population selection strategy based on the residue-residue contact is designed for the purpose of orientation searching.Finally,the probability distribution function is designed to achieve the adaptive transition of the two selection strategies.3.Based on Rosetta platform in Linux operation system,SCDE is implemented in C++.For a given target protein,the fragment library is obtained by Robetta.PSIPRED server is used for predicting the secondary structure.RaptorX-Contact server is used for predicting the residue-residue contact map.Experimental results on a benchmark dataset with 28 proteins show that the proposed method,SCDE,is effective.Furthermore,the prediction results of 4 free model targets in CASP12 is compared with BAKER-SERVER and Zhang-SERVER to verify the effectiveness of SCDE.