支持向量机方法预测离子通道蛋白  

Application of Support Vector Machines in Silico Prediction of Ion Channel Proteins

作　　者：涂白[1] 毕然[1] 

机构地区：[1]华中科技大学生命科学与技术学院,武汉430074

出　　处：《计算机与数字工程》2007年第10期8-10,共3页Computer & Digital Engineering

基　　金：国家自然科学基金项目(编号:90203011)资助

摘　　要：讨论一种基于蛋白质结构域的方法预测离子通道蛋白。通过将蛋白质的结构域转化成为固定长度的向量,使用支持向量机方法进行离子通道蛋白的预测,并将预测结果与线性判别分析以及利用InterPro与GO映射规则进行预测的结果进行了比较。通过留一法交叉验证,取得最好的预测效果,敏感度为95.9%,专一性为98.3%。Ion channel proteins are greatly related with many diseases and play an important role in the drug discovery. To accurately annotate the function of ion channel proteins, we have developed a computational method based on the protein domain composition and support vector machine (SVM) algorithm. While comparing with the linear discriminant method and the method using map rules between InterPro and gene ontology (GO), the SVM approach could get better prediction accuracy, the sensitivity and specificity are 95.9% and 98.3 respectively.

关 键 词：离子通道 结构域 支持向量机 基因本体 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]