Advances in experimental molecular biology techniques are allowing us to elucidate the molecular mechanisms cell behavior with ever-increasing detail. Consequently, it is now routine for individual studies to employ a variety of high-throughput techniques—each of which characterizes the molecular state of the cell at different layers of regulation—including microarrays to assess genome-wide mRNA expression; high-throughput chromatin immunoprecipitation techniques to assess protein–DNA interactions and epigenetic status; and mass spectrometry proteomics and phosphoproteomics to assess protein expression profiles. However, despite the success of these emerging experimental techniques, the potential gains from these advances are generally not fully realized since high-throughput techniques often produce considerably more data than our current ability to adequately organize and analyze, resulting in loss of potentially valuable information. This problem is particularly acute when considering high-throughput time series. In order to address this data integration challenge, a number of systematically curated databases which collate data from different experiments have been established. For example, the Molecular Signatures Database provides a set of annotated gene sets representing a large number of gene expression profiles from a variety of organisms for use with Gene Set Enrichment Analysis (GSEA) (Subramanian et al.
). In addition, a variety of regulatory networks, which encode systems-level molecular interactions, have also been reconstructed, with the aim of dissecting coordinated or synergistic regulatory control of cell behavior. Along with these networks, a range of tools to infer functional subnetworks that ‘connect’ molecular species of interest—and thereby uncover pathways or regulatory modules responsible for changes in cellular state—have also been developed (Berger et al.
; Ulitsky and Shamir, 2007
). However, many interesting cellular phenomena are dynamic and concern the ways in which transitions from one cellular state to another occur and the molecular mechanisms responsible for accompanying changes in cellular phenotype. For this reason, several algorithms have been developed specifically for clustering and interrogating high-dimensional time series (Ma et al.
; Segal et al.
; Wang et al.
). Of particular note is the Gene Expression Dynamics Inspector (GEDI) (Eichler et al.
). GEDI makes animated movies of changing gene expression patterns using dynamically colored self-organizing maps. However, although GEDI is a powerful computational resource, it does not allow interrogation of dynamic patterns of expression against databases of prior knowledge, thus making it difficult to identify potential functional mechanisms responsible for observed dynamic gene expression patterns. However, several approaches have been developed specifically for this purpose. For example, the Short Time series Expression Miner (STEM) uses a correlation-based clustering in combination with gene ontology enrichment analysis to identify potential regulatory mechanisms responsible for expression changes (Ernst and Bar-Joseph, 2006
). However, STEM results are not dynamically visualized as with GEDI but rather are visualized using line graphs and histograms.