Systems Biology Approach for Characterizing the Condition-Specific Usage and Expression of Genes, Gene Products and Functional Modules

Background: Although a great deal is known about how individual genes operate in different processes, our understanding of their coordination and spatio-temporal dynamics in different biological contexts remains limited. The recent explosion in the quantity and specificity of genomic datasets from several well-characterized model systems as well as humans presents a rich opportunity to develop new approaches to address these questions. With measurements that can probe the entire transcriptome, proteome, localizome and interactome, biologists well positioned to understand the organizing principles of biological systems. With this wealth of information, we have the challenge and chance to capture gene expression dynamics and to understand how gene products and functional modules (collections of gene products functioning together for a common biological purpose) are (re)used throughout development. Here we are interested in developing methods for de-convolving this massive amount of data to identify the condition-specific usage of molecular actors and to learn the features that distinguish them.;Results: We used multiple methods for integrating heterogeneous data types to create condition-specific networks of genes and proteins in Drosophila melanogaster, Caenorhabditis elegans, and human lung fibroblasts. These networks revealed interactions and modules specific to our tested conditions. Our study of RNA-seq data from every developmental stage of C. elegans enabled us to discover alternative transcript splice isoforms that were differentially expressed. Additionally, we built Module Identification in Networks (MINE) to perform graph-based clustering and created a conceptual framework for using it in a condition-specific manner. Finally, we applied a principled method for bicluster discovery, cMonkey, to rigorously identify condition-specific clusters (CSCs) and describe their usage in C. elegans..;Conclusions: We found that by combining multiple data types we were able to supplement our knowledge of any particular biological system, and using this aggregate dataset, we were able to discover properties specific to our processes of interest. With cMonkey derived CSCs, we found modules specific to C. elegans tissues, particularly genes that may play a role in the pharynx. Finally, we have outlined a preliminary description of how potential pleiotropic genes can be discovered from genes that are clustered into multiple CSCs.