Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
15542 | Computational Biology and Chemistry | 2007 | 6 Pages |
Finding transcription factors (TFs) to their target genes (TGs) is the first step to understand the transcriptional regulatory networks. Here we present a method which uses an enhanced Bayesian classifier to predict the TF–TG pairs in time-course expression data. Different from previous prediction models, the gene expression data is encoded by discrete values and the temporal feature is used in the enhanced Bayesian classifier. The enhanced Bayesian classifier is trained and tested on two groups of positive and negative samples by three-fold cross-validation and compared with other methods. As the prediction result is improved obviously, the enhanced Bayesian classifier represents a new perspective on studying the regulation relationships from gene expression data. Further more, a data selection method which focus on ‘active’ TFs is proposed, suggesting a new approach on selecting effective time-course expression data.