Identification of Dynamic Metabolic Flux Balance Models Based on Parametric Sensitivity Analysis

Abstract

This paper deals with robust identification of dynamic metabolic flux model parameters based on parametric sensitivity analysis. First, the parameters in the model are ranked based on a global parametric sensitivity analysis to assess whether a subset of the parameters can be eliminated from further analysis. Then, the remaining significant parameters are identified based on the maximization of an overall parametric sensitivity measure subject to set based constraints that are derived from available data. The proposed method is illustrated on a case study of a model describing the diauxic growth of Escherichia coli in a batch culture.

Online Model Maintenance via Output Modifier Adaptation

“…Our proposed method is similar in spirit with the simultaneous identification and optimization method proposed by Martinez Villegas et al and Hille and Budman . In their work, the authors also study on the synergy between identification and optimization via Modifier Adaptation.…”

“…Despite the fact that the model adaptation step is not explicitly related to the first order modifiers, their parameter estimation objective function minimizes the error between the gradients of the model-based optimization problem and the plant, for both the cost and constraint. Note that the objective of the method proposed in our paper and the one proposed by Martinez Villegas et al . and Hille and Budman are related, as they both determine the best model based on minimizing the value of the modifiers.…”

Experimental Design for Batch-to-Batch Optimization under Model-Plant Mismatch

“…The set measurements for all sampling times t i along each batch can then be defined as where the set-based bounds provide upper and lower bounds for the permissible range of model outputs at each sampling time t i such that where the model outputs are given by . Set-based bounds such as defined in eq are particularly well suited for describing experimental data in biological systems , for which the variation among batches typically arises from variability in initial concentrations, changes in environmental conditions, or biases in analytical equipment such as HPLC. The use of such bounds in the current approach is further justified by the need to bound the time trajectories of the variables when using nonconventional model fitting criteria as explained below.…”

Sugar Extraction via Moving-Bed Diffusers in Ethanol Production Industry: Phenomenological Modeling and Finite-Volumes Simulation

“…Model parameters ( k 1 , k 2 , α, and β) are fitted using a genetic algorithm − (GA). In the simulation framework, the implementation of GA in the Python programming language was performed using the DEAP (Distributed Evolutionary Algorithms in Python) package.…”