蛋白-核酸相互作用的支持向量机预测模型

与核酸作用的蛋白质在基因功能许多方面扮演着极其重要的角色,预测蛋白质是否与核酸作用在生物信息学领域受到广泛关注。本文用氨基酸组成、氨基酸物化特性和蛋白质结构等信息作为特征参数,通过支持向量机方法预测了与核酸作用的蛋白质。分别取与rRNA,RNA和DNA作用的3个蛋白质数据集,用SVM训练,筛选最优核函数,优化核函数参数,建立分类判别模型,并用于预测蛋白质是否与核酸作用。结果表明:即使对同源相似性低于40%的蛋白质,通过用10-crossvalidation(交叉验证)方法测试上述3个数据集都分别有93.75%、83.41%、81.85%的预测正确率。用外部测试集测试所得模型分别有93.8%、84.2%、81.9%的预测正确率。在此基础上,我们建立了1个预测蛋白质与核酸是否作用的网上在线软件系统。网址是:http://chemdata.shu.edu.cn/protein_na。

Prediction of nucleic acid-binding proteins using support vector machines

In this work,we integrated SVMs,protein sequence amino acid composition,and associated physicochemical properties into the study of nucleic-acid-binding proteins prediction.We developed the binary classification for rRNA-,RNA-,DNA-binding proteins that play an important role in the control of many cell processes.Each SVM model can be used to predict whether a protein belongs to rRNA-,RNA-,or DNA-hinding protein class.10-crossvalidation was performed on the protein data sets in which the sequences identity was～40%.Test results show that the accuracies of SVM models for rRNA-,RNA-,DNA-binding proteins are 93.75%,83.41%,81.85%,respectively.The predictions were also performed on the test data set.The results agree well with the prior knowledge of those proteins and show the effectiveness of physicochemical properties of sequence in the protein function prediction.On the basis of our work,an online server for predicting the nucleic acid-binding proteins using SVM was available on http://chemdata,shu.edu.cn/protein_na.