The overall goal of this proposed research is to develop advanced techniques for reconstructing transcriptional regulatory networks and deducing the underlying dynamics of transcription factors from gene expression measurements. Specifically, the proposed research will extend the Network Component Analysis (NCA) developed in the sponsor's laboratory to reconstruct network models where limited or incomplete connectivity information of the underlying network is available. Starting with a transcriptional network deduced from DNA binding motif data or other high-throughput experiments, we will develop a multi-focusing strategy for elucidating the unknown functional and indirect connectivity between gene expression and control factors. Some of these control factors include transcription factors, small regulatory molecules in prokaryotes, and chromatin remodeling complexes in eukaryotes. We will apply the multi-focusing algorithm to analyze both Escherichia coli and Saccharomyces cerevisiae microarray data. For E. coli, we will particularly examine the effect of small molecular effectors such as ppGpp at the system-level regulatory network dynamics. For the S. cerevisiae, we will model the effect of chromatin remodeling and TF-TF interaction on gene expression, and construct a quantitative model to describe these functional linkages between genes and control factors (TF or remodeling complexes) within the regulatory network. While the proposed research will focus on studying transcriptional regulatory networks, the computational advances developed is directly applicable to model many different kinds of combinatorial control systems such as metabolic or signal transduction networks, as well as interpreting the state-of-the-art experiments such as fMRl. This wide applicability further increases the impact and significance of the proposed research.
