<i>In silico</i> Metabolic Flux Data Flexibilization for Advanced Bioreactor Control Applications

Mesquita, Thiago José Barbosa; Giordano, Roberto de Campos; Zangirolami, Teresa Cristina; Horta, Antonio Carlos Luperni

doi:10.1089/ind.2020.29197.tjm

Cited by 1 publication

(7 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The proposed micro‐aeration control required a detailed evaluation of the influence of oxygen on yeast metabolism. Therefore, the MM was used to map different combinations of input fluxes (

J_{S}^{MM}

and

J_{{normalO}_{2}}^{MM}

) and the corresponding outputs: ethanol, biomass, CO 2 , and glycerol fluxes to help visualize the best operational region (Mesquita et al, 2020). Regarding the posterior control application, the input/output structure selected for the ANN has the normalized

J_{{CO}_{2}}^{ANN}

(Equation 2) as input and the normalized

J_{{normalO}_{2}}^{ANN}

as output (Figure 4).…”

Section: Resultsmentioning

confidence: 99%

“…The carbon source used was glucose and all the input and output fluxes were compiled in matrices. The matrix profiles of the highest ethanol fluxes (

J_{EtOH}^{MM}

) were selected for ANN training (Mesquita et al, 2019, 2020).…”

Section: Methodsmentioning

confidence: 99%

“…The selected metabolic flux information was normalized and used in the training and validation of multilayer perceptron neural networks (Nelles, 2001). The ANN design followed a simple approach (Horta et al, 2011, 2014; Mesquita et al, 2020). The neural network had a single input‐single output (SISO) configuration

J_{CO 2}^{ANN}

was the input and

J_{normalO 2}^{ANN}

the output.…”

Section: Methodsmentioning

confidence: 99%

“…Another method consists of using surrogate or meta‐models, which are simpler but preserve the same basic responses of a full model (Papadopoulou et al, 2010). These models range from a simple mathematical correlation (Mesquita et al, 2019) to ones generated by more advanced deep learning techniques, such as artificial neural networks (ANNs) (Ibrahim et al, 2019; Mesquita et al, 2020).…”

Section: Introductionmentioning

confidence: 99%

“…For bioprocesses, ANNs can learn complex functional relations among variables (Stanke & Hitzmann, 2012), enabling them to be used as soft sensors (Beiroti et al, 2019; Tavasoli et al, 2019) or surrogate models (Boukouvala et al, 2010; Cozad et al, 2014). So far, there have been few studies concerning the use of ANNs to reproduce the rich information provided by GSMs within control structures (Mesquita et al, 2020).…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Machine learning applied for metabolic flux‐based control of micro‐aerated fermentations in bioreactors

Mesquita

Campani

Giordano

et al. 2021

Biotech & Bioengineering

View full text Add to dashboard Cite

Various bio‐based processes depend on controlled micro‐aerobic conditions to achieve a satisfactory product yield. However, the limiting oxygen concentration varies according to the micro‐organism employed, while for industrial applications, there is no cost‐effective way of measuring it at low levels. This study proposes a machine learning procedure within a metabolic flux‐based control strategy (SUPERSYS_MCU) to address this issue. The control strategy used simulations of a genome‐scale metabolic model to generate a surrogate model in the form of an artificial neural network, to be used in a micro‐aerobic fermentation strategy (MF‐ANN). The meta‐model provided setpoints to the controller, allowing adjustment of the inlet air flow to control the oxygen uptake rate. The strategy was evaluated in micro‐aerobic batch cultures employing industrial Saccharomyces cerevisiae yeast, with defined medium and glucose as the carbon source, as a case study. The performance of the proposed control scheme was compared with a conventional fermentation and with three previously reported micro‐aeration strategies, including respiratory quotient‐based control and constant air flow rate. Due to maintenance of the oxidative balance at the anaerobiosis threshold, the MF‐ANN provided volumetric ethanol productivity of 4.16 g·L−1·h−1 and a yield of 0.48 gethanol.gsubstrate−1, which were higher than the values achieved for the other conditions studied (maximum of 3.4 g·L−1·h−1 and 0.35–0.40 gethanol·gsubstrate−1, respectively). Due to its modular character, the MF‐ANN strategy could be adapted to other micro‐aerated bioprocesses.

show abstract

“…The proposed micro‐aeration control required a detailed evaluation of the influence of oxygen on yeast metabolism. Therefore, the MM was used to map different combinations of input fluxes (

J_{S}^{MM}

and

J_{{normalO}_{2}}^{MM}

J_{{CO}_{2}}^{ANN}

(Equation 2) as input and the normalized

J_{{normalO}_{2}}^{ANN}

as output (Figure 4).…”

Section: Resultsmentioning

confidence: 99%

“…The carbon source used was glucose and all the input and output fluxes were compiled in matrices. The matrix profiles of the highest ethanol fluxes (