Reconstruction and systems analysis of genome-scale metabolic and regulatory networks in Saccharomyces cerevisiae

Modern biological research focuses on determining the structure and function of the biochemical networks operating inside cells. Experimental methods such as genome sequencing or protein-DNA interaction assays that allow determining the components of these networks and mapping their interactions provide the basis for reconstructing network structures. Deciphering how particular network structures give raise to the experimentally measured network states such as gene expression and growth phenotypes is one of the key challenges in biological research. In this dissertation, an approach based on constraint-based analysis of reconstructed genome-scale metabolic and regulatory networks is used to address this question in S. cerevisiae. Genome-scale metabolic and transcriptionally regulated metabolic networks in yeast are reconstructed based on genetic, biochemical, and physiological data. The reconstructed networks are then used as basis for iterative model building in order to improve and expand the network structures. In the case of metabolism, the model is found to correctly predict 82.5% of over 4,700 distinct deletion strain growth phenotypes, and the mispredictions are used to suggest specific changes to the network reconstruction. To extend this work, a computationally efficient method for systematically identifying the optimal changes to a constraint-based metabolic model in order to minimize error between predicted and measured fluxes is developed and applied to data from experimentally evolved E. coli strains. Integrating the reconstructed transcriptional regulatory network with the metabolic network allows studying the interplay between these two networks and analyzing disparate data types such as gene expression and growth phenotypes within the same framework. The integrative analysis is demonstrated using a genome-scale regulated metabolic model to quantitatively predict growth rates of transcription factor deletion strains and using gene expression data for the same deletion strains to identify changes in the regulatory network that result in improved growth rate predictions. Finally, the regulated metabolic model is used to study the effects of transcriptional regulation on the global functional connectivity of the metabolic network. In summary, this dissertation demonstrates the utility of integrated genome-scale models of biochemical networks in data interpretation and network expansion, as well as in understanding the design principles of these networks.