基于接触图残基对距离约束的蛋白质结构预测算法

从头预测是蛋白质结构建模的一种重要方法,该方法的研究有助于人类理解蛋白质功能,从而进行药物设计和疾病治疗。为了提高预测精度,文中提出了基于接触图残基对距离约束的蛋白质结构预测算法(CDPSP)。基于进化算法框架,CDPSP将构象空间采样分为探索和增强两个阶段。在探索阶段,设计基于残基对距离的变异与选择策略,即根据接触图的接触概率选择残基对,并通过片段组装技术对所选择的残基对的邻近区域进行变异;将残基对距离离散化为多个区域并为其分配期望概率,根据期望概率确定是否选择变异的构象,从而增加种群的多样性。在增强阶段,利用基于接触图信息的评分指标,结合能量函数,衡量构象的质量,从而选择较优的构象,达到增强CDPSP近天然态区域采样能力的效果。为了验证所提算法的性能,通过CASP12中的10个FM组目标蛋白质对其进行了测试,并将其与一些先进算法进行比较。实验结果表明,CDPSP可以预测得到精度较高的蛋白质三维结构模型。

Contact Map-based Residue-pair Distances Restrained Protein Structure Prediction Algorithm

De novo prediction is an important method for protein structure modeling.Research of the method contributes to humanity’s understanding of protein functions to conduct drug design and disease treatments.In order to improve the accuracy of prediction,contact map-based residue-pair distances restrained protein structure prediction algorithm(CDPSP)was proposed.Based on the framework of evolutionary algorithm,CDPSP was used to sample conformational space,which was divided into exploration and exploitation stages.In the exploration stage,the strategies of mutation and selection were designed on the basis of the distances of residue-pair,which can increase the diversity of the population.In detail,a residue-pair was chosen according to the contact probability of contact map and the mutation was conducted through fragment assembly technique on the adjacent region of the residue-pair.The selection of mutated conformation was determined by the expected probability distributed through the discretization of residue-pair distances.In the exploitation stage,the contact-based score and energy function were used to evaluate the quality of conformations in search of good conformations,which can enhance the sampling ability of CDPSP in near-native region.In order to verify the performance of the proposed algorithm,CDPSP is tested on 10 targets in the FM group of CASP12 and compared with advanced algorithms.The test results show that CDPSP can predict more accurate protein tertiary structure models.