A systematic approach to reconstructing transcription networks in Saccharomycescerevisiae

Abstract

Decomposing regulatory networks into functional modules is a first step toward deciphering the logical structure of complex networks. We propose a systematic approach to reconstructing transcription modules (defined by a transcription factor and its target genes) and identifying conditionsperturbations under which a particular transcription module is activateddeactivated. Our approach integrates information from regulatory sequences, genome-wide mRNA expression data, and functional annotation. We systematically analyzed gene expression profiling experiments in which the yeast cell was subjected to various environmental or genetic perturbations. We were able to construct transcription modules with high specificity and sensitivity for many transcription factors, and predict the activation of these modules under anticipated as well as unexpected conditions. These findings generate testable hypotheses when combined with existing knowledge on signaling pathways and protein-protein interactions. Correlating the activation of a module to a specific perturbation predicts links in the cell's regulatory networks, and examining coactivated modules suggests specific instances of crosstalk between regulatory pathways.

Fig 1.

(a) A schematic diagram of the intracellular networks. Solid orange arrows represent detection of different cellular conditions, blue dashed arrows represent possible activation of signaling pathways other than the default one, solid cyan arrows represent transcription factors' regulation of their target genes, and the orange dashed line represents possible protein–protein interactions. (b) Decomposition of the intracellular networks into transcription modules. Solid orange and cyan arrows represent detection and transduction of extracellular signals, respectively, and red dashed arrows represent interactions between transcription modules.

Fig 2.

Mitogen-activated protein kinase (MAPK) pathways [modified from Roberts et al. (23)]. Genes, if deleted, to activate or deactivate Ste12p, as inferred from X vector comparison, are correspondingly red or green.