基于多角度进化保守性的蛋白质热区验证方法研究

Being the result of the more advanced research on hot regions of protein-protein interactions, lots of hot regions are predicted by intelligent computing, and it is becoming an important task to validate these hot regions efficiently and precisely; while as a major on-going method, the biological validation based experiments are not practical due to its limited capabilities at high time cost and complexity against large scale of samples. By using protein biological feature-linked machine learning algorithm, and starting from the evolutionary conservations of hot spot residues, this study performs a serial of sequence, structure and energy conservation validations of hot regions, which verify hot regions by using the conservation tendency threshold module adopting the conservation scores of multiple sequence alignments, module substitute matrixes, structural similarities and molecular dynamics simulations. This study has important theoretical and practical contributions to improving the research on protein hot regions, and benefits the protein activities’ understanding of disease origin, biological pharmacy and drug targets.

随着蛋白质相互作用热区研究的不断发展，智能计算产生大量预测热区，高效准确的验证这些预测热区已经成为热区研究的关键问题。现有的方法大多基于生物实验验证，由于实验的长周期性和复杂性，对大规模样本的验证能力非常有限。本项目从热点残基的进化保守性出发，利用机器学习算法结合蛋白质的生化特征，从序列保守性、结构保守性和能量保守性进行多角度验证，通过多序列比对、模块替换矩阵、结构相似性和分子动力学模拟等多方面构建保守性得分算法，利用保守倾向性阈值判定模块对热区进行验证。本项目对于推进蛋白质热区研究，进而对理解疾病的起源、生物制药、药物作用靶点等蛋白质活动，都有重要的理论和实际意义。

本项目从热点残基的进化保守性出发，利用机器学习算法结合蛋白质的生化特征，从序列保守性、结构保守性和能量保守性进行多角度验证，通过多序列比对、模块替换矩阵、结构相似性和分子动力学模拟等多方面构建保守性得分算法，利用保守倾向性阈值判定模块对热区进行验证，最终综合蛋白质序列、结构和能量形成一个完整的蛋白质热区验证模型，该模型对于推进蛋白质热区研究，进而对理解疾病的起源、生物制药、药物作用靶点等蛋白质活动，都有重要的理论和实际意义。