蛋白质相互作用热点残基的预测  

Prediction of Hot Spots at Protein-Protein Interface

作　　者：张少辉[1] 张晓龙[1] 

机构地区：[1]武汉科技大学计算机科学与技术学院,武汉430065

出　　处：《生物物理学报》2013年第2期151-157,共7页Acta Biophysica Sinica

基　　金：国家科学自然基金项目(61273225);生物电子学国家重点实验室2010年度开放课题(2010D11)~~

摘　　要：氨基酸突变扫描实验揭示了在蛋白质相互作用的结合过程中大部分的结合自由能是由极少数热点残基贡献的,通常定义结合自由能变化△△G≥2.0 kcal/mol的蛋白质残基为热点残基。热点残基对蛋白质相互作用具有重要意义。因此,如何有效进行热点残基的预测,仍然是一个研究课题。综合蛋白质氨基酸理化属性的加权疏水性、加权残基接触数、结构属性溶剂可接近面积和残基突出指数等特征,提出利用机器学习支持向量机算法来预测热点残基的方法。所提方法在丙氨酸热力学数据库数据和结合界面数据库选定的数据集上有很好的效果。在一定程度上对以后的研究发展有所帮助。The experimental study of alanine scanning mutagenesis has shown that most of the binding energies in protein-protein interactions are contributed by a few significant residues at the protein-protein interfaces. These residues are called hot spots. Hot spots are typically defined as those residues for which the change of the binding energies （A/kG）~〉2.0 kcal/mol. A new method was proposed by combining protein amino acid physicochemical features and structural features to predict the hot spots using support vector machine based on Weka. With the data extracted from ASEdb and BID respectively, the experimental results showed high accuracy. The simple and definitive features from previous studies were selected, and the method promoted the development of the hot spots prediction in some degree.

关 键 词：蛋白质相互作用 蛋白质热点残基 支持向量机 

分 类 号：Q816[生物学—生物工程]