The development of genome scale metabolic models have been aided by the increasing availability of genome sequences of microorganisms such as Geobacter sulfurreducens, involved in environmentally relevant processes such as the in-situ bioremediation of U(VI). Since microbial activities are the major driving forces for geochemical changes in the sub-surface, understanding of microbial behavior under a given set of conditions can help predict the likely outcome of potential subsurface bioremediation strategies. Hence, a model based lookup table was created to capture the variation in physiology of G. sulfurreducens in response to environmental perturbations. Thermodynamically feasible flux distributions were generated by incorporating thermodynamic constraints in the model. These constraints together with the mass balance constraints formed the thermodynamics based metabolic flux analysis model (TMFA). Metabolomics experiments were performed to determine the concentration of intracellular metabolites. These concentrations were posed as constraints in the TMFA model to improve the model accuracy.

Plasmodium falciparum is the causative agent of malaria, one of the world‟s most prevalent infectious diseases. The emergence of strains resistant to current therapeutics creates the urgent need to identify new classes of antimalarials. Here we present and analyse a constraints-based model (iMPMP427) of P. falciparum metabolism. Consisting of 427 genes, 513 reactions, 457 metabolites, and 5 intracellular compartments, iMPMP427 is relatively streamlined and contains an abundance of transport reactions consistent with P. falciparum’s observed reliance on host nutrients. Flux Balance Analysis simulations reveal the model to be predictive in regards to nutrient transport requirements, amino acid efflux characteristics, and glycolytic flux calculation, which are validated by a wealth of experimental data. Furthermore, enzymes deemed to be essential for parasitic growth by iMPMP427 lend support to several previously computationally hypothesized metabolic drug targets, while discrepancies between essential enzymes and experimentally annotated drug targets highlight areas of malarial metabolism that could benefit from further research.

Plasmodium falciparum is the causative agent of malaria, one of the world‟s most prevalent infectious diseases. The emergence of strains resistant to current therapeutics creates the urgent need to identify new classes of antimalarials. Here we present and analyse a constraints-based model (iMPMP427) of P. falciparum metabolism. Consisting of 427 genes, 513 reactions, 457 metabolites, and 5 intracellular compartments, iMPMP427 is relatively streamlined and contains an abundance of transport reactions consistent with P. falciparum’s observed reliance on host nutrients. Flux Balance Analysis simulations reveal the model to be predictive in regards to nutrient transport requirements, amino acid efflux characteristics, and glycolytic flux calculation, which are validated by a wealth of experimental data. Furthermore, enzymes deemed to be essential for parasitic growth by iMPMP427 lend support to several previously computationally hypothesized metabolic drug targets, while discrepancies between essential enzymes and experimentally annotated drug targets highlight areas of malarial metabolism that could benefit from further research.

The development of genome scale metabolic models have been aided by the increasing availability of genome sequences of microorganisms such as Geobacter sulfurreducens, involved in environmentally relevant processes such as the in-situ bioremediation of U(VI). Since microbial activities are the major driving forces for geochemical changes in the sub-surface, understanding of microbial behavior under a given set of conditions can help predict the likely outcome of potential subsurface bioremediation strategies. Hence, a model based lookup table was created to capture the variation in physiology of G. sulfurreducens in response to environmental perturbations. Thermodynamically feasible flux distributions were generated by incorporating thermodynamic constraints in the model. These constraints together with the mass balance constraints formed the thermodynamics based metabolic flux analysis model (TMFA). Metabolomics experiments were performed to determine the concentration of intracellular metabolites. These concentrations were posed as constraints in the TMFA model to improve the model accuracy.

There is a global need for sustainable, renewable and clean energy sources. Microbial production of hydrogen from renewable carbon sources, biorefinery compounds such as succinic acid or from food and drinks industry waste meets all these criteria. Although it has been studied for several decades, there is still no large scale bio-hydrogen production because the rate and yield of hydrogen production are not high enough to render the process economical. The dependency of biological hydrogen production of incipient light energy is also an important factor affecting economics. In order to improve the prospects of biohydrogen as a renewable and sustainable energy alternative, the genetic and process engineering approaches should be helped and targeted by metabolic engineering tools such as metabolic flux balance analysis. The overall aim of this research was the development of computational metabolic flux balance analysis for the study of growth and hydrogen production in Rhodobacter sphaeroides. The research reported in this thesis had two approaches; experimental and computational. Batch culture experiments for growth and hydrogen production by Rhodobacter sphaeroides were performed with either malate or succinate as carbon source and with glutamate as the nitrogen source. Other conditions investigated included; i) aerobic and anaerobic growth, ii) light and dark fermentation for growth, and iii) continuous light and cycled light/dark conditions for hydrogen production. The best growth was obtained with succinate under anaerobic photoheterotrophic conditions with the maximum specific growth rate of 0.0467 h– 1, which was accompanied with the maximum specific hydrogen production rate of 1.249 mmol(gDW.h)– 1. The range of the photon flux used was 5.457 - 0.080 mmol(gDW.h)– 1. The metabolic flux balance model involved 218 reactions and 176 metabolites. As expected the optimised specific rates of growth and hydrogen production were higher than those of the experimental values. The best prediction was for hydrogen production on succinate with computed specific hydrogen production rates in the range of 2.314 - 1.322 mmol(gDW.h)– 1. Sensitivity analyses indicated that the specific growth rate was affected by the nitrogen source uptake rate under aerobic dark condition whereas the flux of protein formation had the largest effect on the specific growth rate under anaerobic light condition.

Metabolism in cancer cells is adapted to meet the proliferative needs of these cells, with notable changes such as enhanced lactate secretion and glucose uptake rates. In this work, we use the Ensemble Modeling (EM) framework to gain insight and predict potential drug targets for tumor cells. A metabolic network consisting of 58 reactions is considered which accounts for glycolysis, the pentose phosphate pathway, lipid metabolism, amino acid metabolism, and includes allosteric regulation. Experimentally measured metabolite concentrations are used for developing the ensemble of models along with information on established drug targets. The resulting models predicted transaldolase (TALA) and succinate-CoA ligase (SUCOAS1m) to display a significant reduction in growth rate when repressed relative to currently known drug targets. Furthermore, the synergetic repression of transaldolase and glycine hydroxymethyltransferase (GHMT2r) showed a three fold decrease in growth rate compared to the repression of single enzyme targets.

Metabolism in cancer cells is adapted to meet the proliferative needs of these cells, with notable changes such as enhanced lactate secretion and glucose uptake rates. In this work, we use the Ensemble Modeling (EM) framework to gain insight and predict potential drug targets for tumor cells. A metabolic network consisting of 58 reactions is considered which accounts for glycolysis, the pentose phosphate pathway, lipid metabolism, amino acid metabolism, and includes allosteric regulation. Experimentally measured metabolite concentrations are used for developing the ensemble of models along with information on established drug targets. The resulting models predicted transaldolase (TALA) and succinate-CoA ligase (SUCOAS1m) to display a significant reduction in growth rate when repressed relative to currently known drug targets. Furthermore, the synergetic repression of transaldolase and glycine hydroxymethyltransferase (GHMT2r) showed a three fold decrease in growth rate compared to the repression of single enzyme targets.

Constraint-based models of metabolism seldom incorporate capacity constraints on intracellular fluxes due to the lack of experimental data. This can sometimes lead to inaccurate growth phenotype predictions. Meanwhile, other forms of data such as fitness profiling data from growth competition experiments have been demonstrated to contain valuable information for elucidating key aspects of the underlying metabolic network. Hence, the optimal capacity constraint identification (OCCI) algorithm is developed to reconcile constraint-based models of metabolism with fitness profiling data by identifying a set of flux capacity constraints that optimally fits a wide array of strains. OCCI is able to identify capacity constraints with considerable accuracy by matching 1,155 in silico-generated growth rates using a simplified model of Escherichia coli central carbon metabolism. Capacity constraints identified using experimental fitness profiles with OCCI generated novel hypotheses, while integrating thermodynamics-based metabolic flux analysis allowed prediction of metabolite concentrations.

Constraint-based models of metabolism seldom incorporate capacity constraints on intracellular fluxes due to the lack of experimental data. This can sometimes lead to inaccurate growth phenotype predictions. Meanwhile, other forms of data such as fitness profiling data from growth competition experiments have been demonstrated to contain valuable information for elucidating key aspects of the underlying metabolic network. Hence, the optimal capacity constraint identification (OCCI) algorithm is developed to reconcile constraint-based models of metabolism with fitness profiling data by identifying a set of flux capacity constraints that optimally fits a wide array of strains. OCCI is able to identify capacity constraints with considerable accuracy by matching 1,155 in silico-generated growth rates using a simplified model of Escherichia coli central carbon metabolism. Capacity constraints identified using experimental fitness profiles with OCCI generated novel hypotheses, while integrating thermodynamics-based metabolic flux analysis allowed prediction of metabolite concentrations.

An interesting discovery in biology is that most genes in an organism are dispensable. That means these genes have minor effects on survival of the organism in standard laboratory conditions. One explanation of this discovery is that some genes play important roles in specific conditions and are essential genes under those conditions. E. coli is a model organism, which is widely used. It can adapt to many stress conditions, including temperature, pH, osmotic, antibiotic, etc. Underlying mechanisms and associated genes of each stress condition responses are usually different. In our analysis, we combined protein abundance data and mutant conditional fitness data into E. coli constraint-based metabolic models to study conditionally essential metabolic genes under temperature and pH stress conditions. Flux Balance Analysis was employed as the modeling method to analysis these data. We discovered lists of metabolic genes, which are E. coli dispensable genes, but conditionally essential under some stress conditions. Among these conditionally essential genes, atpA in low pH stress and nhaA in high pH stress found experimental evidences from previous studies. Our study provides new conditionally essential gene candidates for biologists to explore stress condition mechanisms.