INFERRING THE REGULATORY INTERACTION MODELS OF TRANSCRIPTION FACTORS IN TRANSCRIPTIONAL REGULATORY NETWORKS

摘要

Living cells are realized by complex gene expression programs that are moderated by regulatorynproteins called transcription factors (TFs). The TFs control the di®erential expression of targetngenes in the context of transcriptional regulatory networks (TRNs), either individually or inngroups. Deciphering the mechanisms of how the TFs control the di®erential expression of antarget gene in a TRN is challenging, especially when multiple TFs collaboratively participate innthe transcriptional regulation. To unravel the roles of the TFs in the regulatory networks, wenmodel the underlying regulatory interactions in terms of the TFu0001target interactions' directionsn(activation or repression) and their corresponding logical roles (necessary and/or su±cient). Wendesign a set of constraints that relate gene expression patterns to regulatory interaction models,nand develop TRIM (Transcriptional Regulatory Interaction Model Inference), a new hiddennMarkov model, to infer the models of TFu0001target interactions in large-scale TRNs of complexnorganisms. Besides, by training TRIM with wild-type time-series gene expression data, thenactivation timepoints of each regulatory module can be obtained. To demonstrate the advantagesnof TRIM, we applied it on yeast TRN to infer the TFu0001target interaction models fornindividual TFs as well as pairs of TFs in collaborative regulatory modules. By comparing withnTF knockout and other gene expression data, we were able to show that the performance ofnTRIM is clearly higher than DREM (the best existing algorithm). In addition, on an individualnArabidopsis binding network, we showed that the target genes' expression correlations can bensigni¯cantly improved by incorporating the TFu0001target regulatory interaction models inferredn†† Corresponding author.nJournal of Bioinformatics and Computational BiologynVol. 10, No. 5 (2012) 1250012 (20 pages)n#.c Imperial College PressnDOI: 10.1142/S0219720012500126n1250012-1nby TRIM into the expression data analysis, which may introduce new knowledge in transcriptionalndynamics and bioactivation.