基于蛋白序列的IDQD算法预测蛋白质相互作用  

Predicting Protein-Protein Interactions by IDQD Based on Sequence Information

作　　者：刘佳[1] 裴智勇[1] 邢永强[1] 蔡禄[1] 

机构地区：[1]内蒙古科技大学数理与生物工程学院,内蒙古包头014010

出　　处：《生物数学学报》2010年第3期501-507,共7页Journal of Biomathematics

基　　金：国家自然科学基金资助项目(60761001)

摘　　要：基于蛋白质序列组分信息,提出一个离散增量结合二次判别分析法(IDQD)预测蛋白质相互作用的模型,对人类蛋白质相互作用进行预测.自洽检验的识别精度达到75.89%,3-fold交叉检验的敏感性和特异性分别为64.22%和64.68%.结果表明IDQD算法可以用于蛋白质相互作用的预测.In this work,based on the compositional characteristic of protein sequences, the method of increment of diversity combined with quadratic discriminant analysis was used to predict human protein-protein interactions.The results showed that the prediction accuracy of the self-consistent test is 75.89%and the sensitivity and the specificity of the 3-fold cross-validation test are 64.22%and 64.68%,which indicate that IDQD method is useful in protein-protein interaction prediction.

关 键 词：蛋白质 相互作用 离散增量 二次判别分析 

分 类 号：Q612[生物学—生物物理学]