摘要
从酶序列出发,提出了一种新的酶的亚类的预测方法。利用物化特性编码和离散增量得到特征向量来表示序列信息,并将这些特征向量输入支持向量机,对酶的6个家族类中各自的亚类进行分类。在Jacknife检验下,氧化还原酶,转移酶,水解酶,裂解酶,异构酶和连接酶中包含的亚类得到的预测精度分别为96.2%,99.2%,99.6%,95.3%,94.4%和97.7%。该方法得到的预测结果优于其它方法。
Based on enzyme sequence,a new method for predicting enzyme subclasses is proposed.By using feature vectors with the Biophysics and biochemistrycharacteristic encoding and increment of diversity to express the information of sequence,these vectors as inputting parameters of the SVM,and the SVM are applied to predict enzyme subclasses of the six main family classes.By the Jackknife test,the overall success rates in identifying the subclasses of oxidoreductase,transferases,hydrolases,lyases,isomerases,and ligases are 96.2%,99.2%,99.6%,95.3%,94.4%and 97.7%,respectively.Success rates of our algorithm are higher than other methods
出处
《内蒙古工业大学学报（自然科学版）》
2011年第1期11-16,共6页
Journal of Inner Mongolia University of Technology：Natural Science Edition
基金
内蒙古自治区高等学校科学研究项目(NJZY08059)
内蒙古自然科学基金资助项目(2009MS0111)
关键词
酶的亚类
物化特性编码
离散增量
支持向量机
Jacknife检验
enzyme subclasses
the Biophysics and biochemistry characteristic encoding
increment of diversity
support vector machine
Jackknife test
