AbstractsBiology & Animal Science

Tese de doutoramento em Bioengenharia The goal of Systems Biology is to understand the complex behavior that emerges from the interaction among the cellular components. Industrial biotechnology is one of the areas of application, where new approaches for metabolic engineering are developed, through the creation of new models and tools for simulation and optimization of the microbial metabolism. Although whole-cell modeling is one of the goals of Systems Biology, so far most models address only one kind of biological network independently. This work explores the integration of di erent kinds of biological networks with a focus on the improvement of simulation of cellular metabolism. The bacterium Escherichia coli is the most well characterized model organism and is used as our case-study. An extensive review of modeling formalisms that have been used in Systems Biology is presented in this work. It includes several formalisms, including Boolean networks, Bayesian networks, Petri nets, process algebras, constraint-based models, di erential equations, rule-based models, interacting state machines, cellular automata and agent-based models. We compare the features provided by these formalisms and classify the most suitable ones for the creation of a common framework for modeling, analysis and simulation of integrated biological networks. Currently, there is a separation between dynamic and constraint-based modeling of metabolism. Dynamic models are based on detailed kinetic reconstructions of central metabolic pathways, whereas constraint-based models are based on genome-scale stoichiometric reconstructions. Here, we explore the gap between both formulations and evaluate how dynamic models can be used to reduce the solution space of constraint-based models in order to eliminate kinetically infeasible solutions. The limitations of both kinds of models are leading to new approaches to build kinetic models at the genome-scale. The generation of kinetic models from stoichiometric reconstructions can be performed within the same framework as a transformation from discrete to continuous Petri nets. However, the size of these networks results in models with a large number of parameters. In this scope, we develop and implement structural reduction methods that adjust the level of detail of metabolic networks without loss of information, which can be applied prior to the kinetic inference to build dynamic models with a smaller number of parameters. In order to account for enzymatic regulation, which is not present in constraint-based models, we propose the utilization of Extended Petri nets. This results in a better sca old for the kinetic inference process. We evaluate the impact of accounting for enzymatic regulation in the simulation of the steady-state phenotype of mutant strains by performing knockouts and adjustment of enzyme expression levels. It can be observed that in some cases the impact is signi cant and may reveal new targets for rational strain design. In summary, we have created a solid framework with a common…