基于改进PCA的蛋白质O-糖基化位点的预测

Prediction of the Protein O-glycosylation by Improved Principal Component Analysis

提出了改进的主成分分析(IPCA)的方法,结合支持向量机(SVM)对蛋白质O-糖基化位点进行预测。IPCA克服了传统主成分分析(PCA)寻找全局主要成分的不足,对类内样本进行加权,在保护局部结构的前提下,消除了变量之间的相关性,提取出具有局部特征的主要成分。然后,在特征空间中用SVM进行分类(预测)。实验结果表明,IPCA+SVM方法是预测糖基化位点行之有效的方法。

To improve the prediction accuracy of O-glycosylation sites,a new method of improved principle component analysis(IPCA)was proposed.At first,next the feature of the original data were extracted by IPCA,IPCA protects the local structure of multimodal data by weighting the data in the same class;then the prediction(classification)was done in feature space by Support Vector Machines(SVM).The results indicate that the performance of IPCA+SVM is viable and effect.