基于物理化学性质优化的蛋白质相互作用预测研究  

Prediction of Protein-Protein Interactions Based on Optimized Physicochemical Properties

作　　者：倪青山[1] 王正志[1] 赵英杰[1] 黎刚果[1] 

机构地区：[1]国防科技大学机电工程与自动化学院,中国湖南长沙410073

出　　处：《生命科学研究》2009年第3期204-208,共5页Life Science Research

基　　金：国家自然科学基金资助项目(60471003)

摘　　要：蛋白质相互作用预测是生物信息学研究的重要问题之一.提出了一种基于物理化学性质优化的蛋白质相互作用预测方法,与现有方法的显著不同就是,并未使用已知的氨基酸残基的物理化学性质,而是通过粒子群算法优化得到有益于相互作用预测的物理化学性质数值.对真实的数据集测试表明,优化得到的物理化学性质比现有的物理化学性质更有益于提高蛋白质相互作用的预测性能,与其它方法相比,也具有一定的优势,说明该方法是一种有效的蛋白质相互作用预测方法.Prediction of protein-protein interactions is one of the most important problems in bioinformatics. New methods have been developed based on optimized physieochemical properties to predict protein-protein interactions. Different from the previously developed methods which used the known physicochemieal properties directly, the proposed method uses optimized physieochemieal properties, which can improve the predictive ability of protein-protein interactions and can be obtained by using particle swarm optimization algorithm. Results obtained from the real data sets demonstrated that the optimized physicochemical properties were more beneficial to the prediction of protein-protein interactions than the known ones. Moreover, the proposed method showed better performance than some of the previously developed methods in present research, which indicated that the proposed method may be useful for predicting protein-protein interactions.

关 键 词：蛋白质相互作用 物理化学性质 局部超平面分类算法 粒子群优化算法 

分 类 号：Q332[生物学—遗传学]