Robust Parameter Identification to Perform the Modeling of pta and poxB Genes Deletion Effect on Escherichia Coli.

The aim of this study was to design a robust parameter identification algorithm to characterize the effect of gene deletion on Escherichia coli (E. coli) MG1655. Two genes (pta and poxB) in the competitive pathways were deleted from this microorganism to inhibit pyruvate consumption. This condition deviated the E. coli metabolism toward the Krebs cycle. As a consequence, the biomass, substrate (glucose), lactic, and acetate acids as well as ethanol concentrations were modified. A hybrid model was proposed to consider the effect of gene deletion on the metabolism of E. coli. The model parameters were estimated by the application of a least mean square method based on the instrument variable technique. To evaluate the parametric identifier method, a set of robust exact differentiators, based on the super-twisting algorithm, was implemented. The hybrid model was successfully characterized by the parameters obtained from experimental information of E. coli MG1655. The significant difference between parameters obtained with wild-type strain and the modified (with deleted genes) justifies the application of the parametric identification algorithm. This characterization can be used to optimize the production of different byproducts of commercial interest.

Characterization of nitrogen substrate limitation on Escherichia coli's growth by parameter identification tools.

Carbon-to-nitrogen ratio (CNR) has shown to be a relevant factor in microorganisms growth and metabolites production. It is usual that this factor compromises the productivity yield of different microorganisms. However, CNR has been rarely modeled and therefore the nature of its specific influence on metabolites production has not been understood clearly. This paper describes a parametric characterization of the CNR effect on the Escherichia coli metabolism. A set of parameters was proposed to introduce a mathematical model that considers the biomass, substrate and several byproducts dynamical behavior under batch regimen and CNR influence. Identification algorithm used to calculate the parameters considers a novel least mean square strategy that formalizes the CNR influence in E. coli metabolism. This scheme produced a step-by-step method that was suitable for obtaining the set of parameters that describes the model. This method was evaluated under two scenarios: (a) using the data from a set of numerical simulations where the model was tested under the presence of artificial noises and (b) the information obtained from a set of experiments under different CNR. In both cases, a leave-one-experiment-out cross-validation study was considered to evaluate the model prediction capabilities. Feasibility of the parametric identification method was proven in both considered scenarios.