基于多特征尺度的大肠杆菌启动子预测  

Prediction of the E.coli-K12 Promoter Based on Multi-Feature Selection

作　　者：杨乌日吐[1] 林昊[2] 

机构地区：[1]内蒙古大学研究生工作部,内蒙古呼和浩特010021 [2]电子科技大学生命科学与技术学院,四川成都610054

出　　处：《计算机工程与科学》2010年第9期148-151,共4页Computer Engineering & Science

基　　金：内蒙古大学青年科学基金资助项目(ND0810);电子科技大学科研启动费资助项目

摘　　要：本文对实验证实的741条大肠杆菌Sigma70启动子的序列进行预测研究。首先,基于RNA聚合酶与DNA的相互作用,利用位置打分函数对序列中的保守位点进行了衡量;然后,根据启动子的序列特征,利用离散性指标对序列中不同的碱基信息含量进行测量;最后,利用多元非线性判别分析实现了对大肠杆菌启动子的预测。10折叠交叉检验结果显示,总体预测精度达到85%以上。与其它算法比较结果显示,我们开发的这一算法能够更好地预测大肠杆菌启动子。According to the known knowledge of 741 experimentally confirmed Sigma70 promoters,the promoters are predicted. At first,based on the interaction between RNAp and DNA elements,the position-correlation-score-function （PCSF） algorithm is used to measure the conservative sits in promoter sequences. Subsequently,according to the characteristics of promoters,a diversity index is applied to measure the information content in different regions. Finally,the modified Mahalanobis Discriminant is proposed to perform prediction. The overall accuracies of 10-fold cross-validation of 85%＋ are achieved. By comparing with other methods,it is shown that the proposed method can recognize the Escherichia coli promoters with high accuracy.

关 键 词：大肠杆菌启动子 位置关联打分函数 多样性指标 修正的马氏判别式 

分 类 号：Q61[生物学—生物物理学]