Lawrence Berkeley National Laboratory

One of the major challenges for the bioinformatics community in view of constantly growing number of complete genomes is providing effective tools to enable high-quality reconstruction of transcriptional regulatory networks (TRN). Definition of a particular TRN includes specification of which transcription factors (TF) bind to TF-binding sites (TFBS) in the promoter regions of which genes and what is the integrated effect of all these TFs on the expression of al these genes. Reconstruction of TRNs helps to better understand the metabolism and functions of bacteria. Among different approaches that are used for TRN reconstruction are an expression data-driven approach, and comparative genomic approaches that are either computing-driven, or subsystem (pathway) -driven. DNA microarrays, reporting gene expression, continue to be an important tool for high-throughput measurements on transcriptional levels, and machine-learning approaches were used to identify TRN (without a TFBS component) from a compendium of microarray expression profiles . However, in many cases the complexity of the interactions between regulons makes it difficult to distinguish between direct and indirect effects on transcription. Availability of a large number of complete genomes opens an opportunity to apply modern approaches of comparative genomics to expand the known regulons to yet uncharacterized organisms and to predict and describe new regulons with high precision.