基于序列信息的酶的亚类多特征参数识别方法

Recognition of Enzyme Subclasses by Multiple Characteristic Parameters Based on Sequence Information

基于酶的序列信息,分别使用矩阵打分与离散增量的方法提取各类特征参数进行有效组合,利用支持向量机分类算法对数据集中酶家族类的各个亚类进行分类识别,获得了最佳的预测结果。在刀切法(Jackknife)检验下,氧化还原酶、转移酶、水解酶、裂合酶、异构酶和合成酶中亚类的总体预测成功率分别为96.43%、92.90%、90.85%、99.22%、99.84%和98.86%。预测结果表明,多特征参数的支持向量机方法明显优于单特征参数的矩阵打分方法和离散增量方法,可以有效识别酶家族类中的亚类。

Based on sequence information of enzyme,matrix scoring and increment of diversity were used to extract various characteristic parameters for effective combination.Support vector machine(SVM)classification algorithm was used to classify and recognize each subclass of enzyme family in the data set,and the best prediction results were obtained.The overall jackknife prediction success rates of the subclasses of oxidoreductase,transferase,hydrolase,lyase,isomerase and ligase are 96.43%,92.90%,90.85%,99.22%,99.84%and 98.86%,respectively.The prediction results show that the SVM method with multiple characteristic parameters is superior to the matrix scoring and increment of diversity with single characteristic parameter,and can effectively identify enzyme subclasses.