Using Computations to Analyze and Redesign Metabolism

Open Access
Ranganathan, Sridhar
Graduate Program:
Integrative Biosciences
Doctor of Philosophy
Document Type:
Date of Defense:
December 06, 2011
Committee Members:
  • Costas D Maranas, Dissertation Advisor
  • Costas D Maranas, Committee Chair
  • Reka Z Albert, Committee Member
  • Eric Thomas Harvill, Committee Member
  • Howard M Salis, Committee Member
  • Cooduvalli S Shashikant, Committee Member
  • Metabolic Engineering
  • Optimization
With the availability genome-wide datasets for various organisms, research in biology has moved towards a systems-level analysis that portrays a comprehensive picture of cellular physiology. In the recent past, complete inventories of all the known genetic capabilities of microorganisms have been built into computational models. In this regard, we present studies aimed at making use of computational models of metabolism and developing computational algorithms that help in analyzing and redesigning the metabolism of microorganisms. We use procedures developed in this thesis as a predictive tool for exploring genetic manipulations that lead to the overproduction of value-added biochemicals. The key aims addressed in this work are - Aim 1: to assemble novel biosynthesis routes from a known substrate or starting metabolite to biofuel molecules, Aim 2: to develop computational / mathematical procedures that suggest genetic interventions in microorganisms that lead to the overproduction of value-added biochemicals. Throughout this work, we rely on systems biology, graph-theory and novel optimization approaches (i.e. bilevel optimization) to address problems related to Aims 1 and 2 in metabolic engineering. The potential of genetically engineering “user-friendly” microbes with non-native biosynthetic gene cascades has been identified as a promising method to produce biofuels during fermentation. To address this problem in Aim 1, we introduce a graph-based pathway prospecting approach that can help uncover all possible metabolic pathways to biofuel molecules such as 1-butanol. We derive computational predictions for these pathways by culling information from databases such as BRENDA, and KEGG. Subsequently under Aim 2, we introduce a novel computational strain redesign procedure called OptForce custom-made to predict genetic interventions (i.e. up-/down-regulations, knockouts, knock-ins) that guarantees a pre-specified yield for the target biofuel molecule. In addition to its capability of predicting multiple genetic interventions at a time, OptForce is also primed to incorporate experimental data (i.e. metabolic flux analysis data) within the procedure before starting the procedure for redesign. In this work, we demonstrate the validity of OptForce by comparing computational predictions with metabolic engineering experiments for overproducing various biochemicals such as succinate, 1-butanol (from the pathways identified under Aim 1), flavanones, fatty and amino acids in Escherichia coli.