基于组合向量的支持向量机算法预测酶的类型

Prediction of Enzyme Types by Support Vector Machine Algorithm Based on Combined Vectors

酶的类型与其功能和催化性能关系密切,对于新发现的酶,可以通过确定它的类型来表明其生物功能。目前研究酶的生化实验方法不仅费时、耗资,而且可能会碰到许多无法解决的实际困难,因此使用机器学习算法来识别酶类型的理论方法显得十分重要。从酶的一级序列出发,给截取的氨基酸片段打分,将打分值和离散增量值构成的组合向量作为特征参数,运用支持向量机的算法预测酶的类型,取得了较好的预测结果,Jackknife检验的预测成功率为88.86%,表明此算法对于酶的分类预测非常有效。

The type of enzyme is closely related to its function and catalytic performance.For a newly discovered enzyme,its biological function can be indicated by determining its type.At present,the biochemical experimental methods for studying enzymes are not only time-consuming and costly,but also may encounter many practical difficulties that cannot be solved.Therefore,the theoretical method for using machine learning algorithm to identify enzyme types is very important.The amino acid sequences of enzymes are scored for specific amino acid segments.The combined vector composed of scoring value and discrete increment value is used as information parameter to predict the type of enzyme by using the algorithm of support vector machine.The prediction success rate of Jackknife test is 88.86%.The results show that this algorithm is very effective for prediction of enzymes.