基于分段氨基酸组成成分的蛋白质相互作用预测

蛋白质相互作用研究有助于揭示生命过程的许多本质问题，也有助于疾病预防、诊断，对药物研制具有重要的参考价值。文章首先构建出蛋白质作用数据库，提出分段氨基酸组成成分特征提取方法来预测蛋白质相互作用。10CV检验下，基于支持向量机的3段氨基酸组成成分特征提取方法的预测总精度为86.2%，比传统的氨基酸组成成分方法提高2.31个百分点；采用Guo的数据库和检验方法，3段氨基酸组成成分特征提取方法的预测总精度为90.11%，比Guo的自相关函数特征提取方法提高2.75个百分点，从而表明分段氨基酸组成成分特征提取方法可有效地应用于蛋白质相互作用预测。

Predicting protein-protein interaction based on the sequence-segmented amino acid composition

The research on protein-protein interaction (PPI) can help us to reveal many essential problems of life processes, and also administer to prevention and diagnosis of human's diseases. It has important reference value for drug development. A dataset of protein-protein interactions was constructed firstly, and then the feature extracting method of sequence-segmented amino acid composition (SAAC) was proposed to predict protein-protein interaction in this paper. Based on the support vector machine (SVM), the prediction accuracy of 3 segments SAAC is 86.2% in 10-fold cross-validation (10CV) test, which is 2.31% higher than that of common amino acid composition (AAC) method. Using Guo's database and test method, the prediction accuracy of 3 segments SAAC is 90.11%, which is 2.75% higher than that of Guo's approach. The results show that the SAAC method can predict protein-protein interaction effectively.