基于递归模糊神经网络的多时延基因调控网络构建方法  

An approach to construct time-lagged gene regulation network based on recurrent fuzzy neural network

作　　者：徐赛娟[1] 郭红[1] 

机构地区：[1]福州大学数学与计算机科学学院,福建福州350108

出　　处：《福州大学学报（自然科学版）》2012年第2期165-171,共7页Journal of Fuzhou University(Natural Science Edition)

基　　金：福建省自然科学基金资助项目(2009J01283)

摘　　要：为了处理连续的时序基因表达数据,提出了一个基于递归模糊神经网络的多时延基因调控网络构建算法.该算法能直接用于分析连续的时序基因表达数据,避免离散化数据造成的信息损失.使用时序互信息估计基因间的转录时延,并限制每个基因的潜在调控基因,从而有效提高建网的效率和精确度.酵母菌细胞周期表达数据的实验结果表明该算法能正确选择潜在的调控基因,更加精确地构建基因调控网络.In order to deal with continuous time series gene expression data, this paper presents an al- gorithm to construct time - lagged gene regulation network based on recurrent fuzzy neural network. The algorithm can he directly used to analyze continuous time series gene expression data, thus to a- void information loss caused by data discretization. In addition to using time series mutual information to estimate transcription delay between genes, the algorithm also limits potential regulation genes of each gene, thus effectively improves the efficiency and accuracy of constructed network. Experimental results on yeast cell cycle expression data show that the tion genes, and more accurately construct gene regulati algorithm can correctly select potential on network. regula

关 键 词：基因调控网络 递归模糊神经网络 时延 时序互信息 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]