以位置特异性得分矩阵和基因本体为特征的蛋白质亚细胞定位预测

Predicting subcellular localization of protein based on PSSM and GO features

提出一种蛋白质亚细胞定位预测方法.该方法以位置特异性得分矩阵和基因本体抽取对应特征,结合支持向量机构建多标签分类模型.充分考虑了蛋白质进化信息对其亚细胞定位的影响,并基于文本分类中涉及到的卡方检验的对数变换思想,构建基因本体注释信息的加权系数对其进行加权处理,从而提高预测的准确率.采用支持向量机作为基分类器构建多标签分类模型,进一步提高预测的准确率.通过在目前该领域两个常用的真实数据集上进行的一系列测试结果表明,该方法能有效提高蛋白质亚细胞定位预测的准确率.

This paper puts forward a computational method for the problem of predicting protein subcellular localization. It uses features of the position specific scoring matrix （PSSM） and gene ontology （GO）, and uses the support vector machine （SVM） to construct a classifier. This method considers the effect of evolutionary information to the subcellular location and also employs text mining method a logarithmic transformation of CHI to determine weight for GO features to achieve higher precision. Furthermore, using SVM as base classifier to construct multi-label classification model, it is performed on two benchmark datasets. The application result shows that the proposed method effectively improves the accuracy of prediction of protein subcellular localization.