期刊
JOURNAL OF BIOMOLECULAR STRUCTURE & DYNAMICS
卷 31, 期 1, 页码 115-124出版社
TAYLOR & FRANCIS INC
DOI: 10.1080/07391102.2012.691368
关键词
microarray; data analysis; heuristic learning; regulon; modulon; transcriptional factors
资金
- Russian Fund of Basic Research [10-04-92,663-IND-a]
- Department of Science and Technology, India [INT/RFBR/P-31]
Identification of bacterial modulons from series of gene expression measurements on microarrays is a principal problem, especially relevant for inadequately studied but practically important species. Usage of a priori information on regulatory interactions helps to evaluate parameters for regulatory subnetwork inference. We suggest a procedure for modulon construction where a seed regulon is iteratively updated with genes having expression patterns similar to those for regulon member genes. A set of genes essential for a regulon is used to control modulon updating. Essential genes for a regulon were selected as a subset of regulon genes highly related by different measures to each other. Using Escherichia coli as a model, we studied how modulon identification depends on the data, including the microarray experiments set, the adopted relevance measure and the regulon itself. We have found that results of modulon identification are highly dependent on all parameters studied and thus the resulting modulon varies substantially depending on the identification procedure. Yet, modulons that were identified correctly displayed higher stability during iterations, which allows developing a procedure for reliable modulon identification in the case of less studied species where the known regulatory interactions are sparse.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据