Using fuzzy logic to design fermentation media: A comparison to neural networks and factorial design

Summary In analysis of oil accumulation by the yeast Rhodotorula gracilis, fuzzy logic and neural networks were shown to be able to significantly reduce the number of experiments required in designing fermentation media. Fuzzy logic performed similarly, although slightly less accurately, than neural networks in predicting the outcome of shake flask experiments. In some instances having too many fuzzy rules decreased the accuracy of the technique, and further work is required to determine the criteria for achieving optimum performance from the fuzzy logic model. Overall fuzzy logic is a viable and useful tool for designing fermentation media.     

Optimization of a fermentation medium using neural networks and genetic algorithms

Artificial neural networks and genetic algorithms are used to model and optimize a fermentation medium for the production of the enzyme hydantoinase by Agrobacterium radiobacter. Experimental data reported in the literature were used to build two neural network models. The concentrations of four medium components served as inputs to the neural network models, and hydantoinase or cell concentration served as a single output of each model. Genetic algorithms were used to optimize the input space of the neural network models to find the optimum settings for maximum enzyme and cell production. Using this procedure, two artificial intelligence techniques have been effectively integrated to create a powerful tool for process modeling and optimization.     

An evaluation of a neural network for the start-up phase of a continuous recombinant fermentation subject to disturbances

Summary A radial basis neural network was applied to a process for glyceraldehyde-3-phosphate dehydrogenase produced by an Escherichia coli strain containing the plasmid pBR Eco gap. A neural network trained with a pure culture predicted the performance of a fermentation containing wild type cells and/or product in the inoculum better than in the reverse case; this is explained. In general, the network learnt the trends in the concentrations of plasmid-containing cells and the recombinant product more accurately than those of wild type cells and the substrate. This similarity with deterministic networks and the good predictability with some training vectors suggests that neural networks can be used to simulate the start-up phase of recombinant fermentations corrupted by disturbances.     

Modeling of the pyruvate production with Escherichia coli: comparison of mechanistic and neural networks-based models

Three different models: the unstructured mechanistic black-box model, the input–output neural network-based model and the externally recurrent neural network model were used to describe the pyruvate production process from glucose and acetate using the genetically modified Escherichia coli YYC202 ldhA::Kan strain. The experimental data were used from the recently described batch and fed-batch experiments [ Zelić B, Study of the process development for Escherichia coli-based pyruvate production. PhD Thesis, University of Zagreb, Faculty of Chemical Engineering and Technology, Zagreb, Croatia, July 2003. (In English); Zelić et al. Bioproc Biosyst Eng 26:249–258 (2004); Zelić et al. Eng Life Sci 3:299–305 (2003); Zelić et al Biotechnol Bioeng 85:638–646 (2004)]. The neural networks were built out of the experimental data obtained in the fed-batch pyruvate production experiments with the constant glucose feed rate. The model validation was performed using the experimental results obtained from the batch and fed-batch pyruvate production experiments with the constant acetate feed rate. Dynamics of the substrate and product concentration changes was estimated using two neural network-based models for biomass and pyruvate. It was shown that neural networks could be used for the modeling of complex microbial fermentation processes, even in conditions in which mechanistic unstructured models cannot be applied.     

Neural network designs for poly- β-hydroxybutyrate production optimization under simulated industrial conditions

Improvement of the fermentation efficiency of poly--hydroxybutyrate (PHB) may make it competitive with chemically synthesized petroleum-based polymers. One step toward this is optimization of fluid dispersion and the feed rates to a fed-batch bioreactor. In a recent study using a fermentation model, dispersion corresponding to a Peclet number of 20 was shown to maximize the productivity of PHB. Here further improvement has been investigated using neural optimization. A comparison of seven neural topologies has shown that while feed-forward and radial basis neural networks are computationally efficient, recurrent networks generate higher concentrations of PHB. All networks enhanced the productivity by 16–93% over model-based optimization.     

Use of uniform designs in combination with neural networks for viral infection process development

This work aimed to compare the predictive capacity of empirical models, based on the uniform design utilization combined to artificial neural networks with respect to classical factorial designs in bioprocess, using as example the rabies virus replication in BHK‐21 cells. The viral infection process parameters under study were temperature (34°C, 37°C), multiplicity of infection (0.04, 0.07, 0.1), times of infection, and harvest (24, 48, 72 hours) and the monitored output parameter was viral production. A multilevel factorial experimental design was performed for the study of this system. Fractions of this experimental approach (18, 24, 30, 36 and 42 runs), defined according uniform designs, were used as alternative for modelling through artificial neural network and thereafter an output variable optimization was carried out by means of genetic algorithm methodology. Model prediction capacities for all uniform design approaches under study were better than that found for classical factorial design approach. It was demonstrated that uniform design in combination with artificial neural network could be an efficient experimental approach for modelling complex bioprocess like viral production. For the present study case, 67% of experimental resources were saved when compared to a classical factorial design approach. In the near future, this strategy could replace the established factorial designs used in the bioprocess development activities performed within biopharmaceutical organizations because of the improvements gained in the economics of experimentation that do not sacrifice the quality of decisions. © 2015 American Institute of Chemical Engineers Biotechnol. Prog., 31:532–540, 2015     

Backpropagation neural network predictive control and control scheme comparison of 2,3-butanediol fermentation by Klebsiella oxytoca

A backpropagation neural network (BPN) was applied for the control study of 2,3-butanediol fermentation (2,3-BDL) carried by Klebsiella oxytoca. The measurements of cell mass and glucose were not included in the network models, instead, only the on-line measured product concentrations from the MIMS (membrane introduction mass spectrometer) were involved. Oxygen composition was chosen to be the control variable for this fermentation system for the formation of 2,3-BDL is regulated by oxygen. Oxygen composition was directly correlated to the measured product concentrations. A two-dimensional (number of input nodes by number of data sets) moving window to supply data for on-line, dynamic learning of this fermentation system was applied. The input nodes of the networks were also properly selected. Two neural network control schemes for this 2,3-BDL fermentation were discussed and compared in this work. Fermentations often exist time delay due to the measurement and their slow reaction nature. Hence, the order of time delay for the network controller was also investigated.     

Application of neural networks for controlling and predicting quality parameters in beer fermentation

The biochemical pathways involved in the production of ethyl caproate, a secondary product of the beer fermentation process, are not well established. Hence, there are no phenomenological models available to control and predict the production of this particular compound as with other related products. In this work, neural networks have been used to fit experimental results with constant and variable pH, giving a good fit of laboratory and industrial scale data. The results at constant pH were also used to predict results at variable pH. Finally, the application of neural networks obtained from laboratory experiments gave excellent predictions of results in industrial breweries and so could be used in the control of industrial operations. The input pattern to the neural network included the accumulated fermentation time, cell dry weight, consumption of sugars and aminoacids and, in some cases, the pH. The output from the neural network was an estimation of quantity of the ethyl caproate ester.     

Improvement of laccase production from Ganoderma sp. KU-Alk4 by medium engineering

To engineer the production of laccase by Ganoderma sp. KU-Alk4, a newly isolated white-rot fungus, a seven-level Box−Behnken factorial design was employed to optimize the culture medium composition. A mathematical model was developed to show the effect of each medium component and their interactions on the production of laccase activity in submerged fermentation. The model estimated the optimal concentrations of glycerol, yeast extract and veratryl alcohol as 40, 0.22 g/l and 0.85 mM, respectively, with the medium pH of 6.0. These predicted conditions were verified by validation experiments. The optimized medium gave laccase activity of 240 U/ml, which is 12 times higher than that produced in non-optimized medium. Thus, this statistical approach enabled rapid identification and integration of key medium parameters for Ganoderma sp. KU-Alk4, resulted the high laccase production.     

Efficient protein crystallization

High-throughput molecular biology and crystallography advances have placed an increasing demand on crystallization, the one remaining bottleneck in macromolecular crystallography. This paper describes three experimental approaches, an incomplete factorial crystallization screen, a high-throughput nanoliter crystallization system, and the use of a neural net to predict crystallization conditions via a small sample (approximately 0.1%) of screening results. The use of these technologies has the potential to reduce time and sample requirements. Initial experimental results indicate that the incomplete factorial design detects initial crystallization conditions not previously discovered using commercial screens. This may be due to the ability of the incomplete factorial screen to sample a broader portion of "crystallization space," using a multidimensional set of components, concentrations, and physical conditions. The incomplete factorial screen is complemented by a neural network program used to model crystallization. This capability is used to help predict new crystallization conditions. An automated, nanoliter crystallization system, with a throughput of up to 400 conditions/h in 40-nl droplets (total volume), accommodates microbatch or traditional "sitting-drop" vapor diffusion experiments. The goal of this research is to develop a fully-automated high-throughput crystallization system that integrates incomplete factorial screen and neural net capabilities.     

Product formation kinetics in genetically modified <Emphasis Type="Italic">E. coli </Emphasis>bacteria: inclusion body formation

A data-driven model is presented that can serve two important purposes. First, the specific growth rate and the specific product formation rate are determined as a function of time and thus the dependency of the specific product formation rate from the specific biomass growth rate. The results appear in form of trained artificial neural networks from which concrete values can easily be computed. The second purpose is using these results for online estimation of current values for the most important state variables of the fermentation process. One only needs online data of the total carbon dioxide production rate (tCPR) produced and an initial value x of the biomass, i.e., the size of the inoculum, for model evaluation. Hence, given the inoculum size and online values of tCPR, the model can directly be employed as a softsensor for the actual value of the biomass, the product mass as well as the specific biomass growth rate and the specific product formation rate. In this paper the method is applied to fermentation experiments on the laboratory scale with an E. coli strain producing a recombinant protein that appears in form of inclusion bodies within the cells’ cytoplasm.     

Application of statistical experimental designs in computer simulations of run-away fermentations

Sets of simulations of run-away fermentations were structured as two-level factorial experimental designs with parameters in a mathematical model as factors. By this technique it was possible to document the robustness and sensitivity of the model and to show why run-away fermentations may be difficult to control in practice. In an engineering approach to fermentation development it is beneficial to integrate simulation of fermentation experiments with real fermentation experiments in order to get a better planning and interpretation.     

Optimization of recombinant aminolevulinate synthase production in<Emphasis Type="Italic"> Escherichia coli</Emphasis> using factorial design

The production of recombinant Rhodobacter sphaeroides aminolevulinate (ALA) synthase was optimized in two strains of Escherichia coli: the wild-type strain MG1655, and a ptsG mutant AFP111. The effects of initial succinate, glucose and isopropyl--d-thiogalactopyranoside (IPTG) concentrations and the time of induction on enzyme activity were studied. One-way analysis was used to approximate the optimal ranges for these factors, followed by a full factorial design to quantify the effects of each factor and the interactions between the factors. Initial succinate, glucose, and IPTG concentration were observed to be the key factors affecting ALA synthase activity with the optimal levels determined to be above 6 g/l succinate, 0 g/l glucose, and 0.10 mM IPTG. ALA synthase activity was generally lower with AFP111 than with MG1655, and the effect of these three key factors was also lower with AFP111 than with MG1655. Based on the full factorial design results, a fermentation was completed that yielded 296 mU/mg protein with a final ALA concentration of 5.2 g/l (39 mM).     

Using historical data for bioprocess optimization: modeling wine characteristics using artificial neural networks and archived process information

Optimization of fermentation processes is a difficult task that relies on an understanding of the complex effects of processing inputs on productivity and quality outputs. Because of the complexity of these biological systems, traditional optimization methods utilizing mathematical models and statistically designed experiments are less effective, especially on a production scale. At the same time, information is being collected on a regular basis during the course of normal manufacturing and process development that is rarely fully utilized. We are developing an optimization method in which historical process data is used to train an artificial neural network for correlation of processing inputs and outputs. Subsequently, an optimization routine is used in conjunction with the trained neural network to find optimal processing conditions given the desired product characteristics and any constraints on inputs. Wine processing is being used as a case study for this work. Using data from wine produced in our pilot winery over the past 3 years, we have demonstrated that trained neural networks can be used successfully to predict the yeast-fermentation kinetics, as well as chemical and sensory properties of the finished wine, based solely on the properties of the grapes and the intended processing. To accomplish this, a hybrid neural network training method, Stop Training with Validation (STV), has been developed to find the most desirable neural network architecture and training level. As industrial historical data will not be evenly spaced over the entire possible search space, we have also investigated the ability of the trained neural networks to interpolate and extrapolate with data not used during training. Because a company will utilize its own existing process data for this method, the result of this work will be a general fermentation optimization method that can be applied to fermentation processes to improve quality and productivity.     

Optimization of carotenoid production by <Emphasis Type="Italic">Rhodotorula glutinis</Emphasis> using statistical experimental design

A two-step optimization strategy of statistical experimental design was employed to enhance carotenoid production from sugar cane molasses (SCM) in the yeast Rhodotorula glutinis. In the first step, a fractional factorial design was used to evaluate the impact of five fermentation factors (pH and concentrations of SCM, urea, KH₂PO₄, and NaCl). The results revealed that three factors (concentrations of SCM, urea, and KH₂PO₄) had a significant influence on biomass and carotenoid production. A face-centered central composite design was then used in the second step to optimize the three significant factors to further enhance the biomass yield and carotenoid production. Through this two-step optimization strategy, the carotenoid concentration could be increased from an average of 1.39 mg/l to an average of 3.46 mg/l, representing a 2.5-fold carotenoid production enhancement.     

Enhanced ethanol fermentation of brewery wastewater using the genetically modified strain <Emphasis Type="Italic">E. coli </Emphasis>KO11

We have used liquid waste obtained from a beer brewery process to produce ethanol. To increase the productivity, genetically modified organism, Escherichia coli KO11, was used for ethanol fermentation. Yeast was also used to produce ethanol from the same feed stock, and the ethanol production rates and resulting concentrations of sugars and ethanol were compared with those of KO11. In the experiments, first the raw wastewater was directly fermented using two strains with no saccharification enzymes added. Then, commercial enzymes, α-amylase, pectinase, or a combination of both, were used for simultaneous saccharification and fermentation, and the results were compared with those of the no-enzyme experiments for KO11 and yeast. Under the given conditions with or without the enzymes, yeast produced ethanol more rapidly than E. coli KO11, but the final ethanol concentrations were almost the same. For both yeast and KO11, the enzymes were observed to enhance the ethanol yields by 61–84% as compared to the fermentation without enzymes. The combination of the two enzymes increased ethanol production the most for the both strains. The advantages of using KO11 were not demonstrated clearly as compared to the yeast fermentation results.     

Modeling of recycling oxic and anoxic treatment system for swine wastewater using neural networks

A recycling reactor system operated under sequential anoxic and oxic conditions for the treatment of swine wastewater has been developed, in which piggery slurry is fermentatively and aerobically treated and then part of the effluent is recycled to the pigsty. This system significantly removes offensive smells (at both the pigsty and the treatment plant), BOD and others, and may be cost effective for small-scale farms. The most dominant heterotrophic were, in order,Alcaligenes faecalis, Brevundimonas diminuta andStreptococcus sp., while lactic acid bacteria were dominantly observed in the anoxic tank. We propose a novel monitoring system for a recycling piggery slurry treatment system through the use of neural networks. In this study, we tried to model the treatment process for each tank in the system (influent, fermentation, aeration, first sedimentation and fourth sedimentation tanks) based upon the population densities of the heterotrophic and lactic acid bacteria. Principal component analysis (PCA) was first applied to identify a relationship between input and output. The input would be microbial densities and the treatment parameters, such as population densities of heterotrophic and lactic acid bacteria, suspended solids (SS), COD, NH₄ ⁺-N, or-tho-phosphorus (o-P), and total-phosphorus (T-P). Then multi-layer neural networks were employed to model the treatment process for each tank. PCA filtration of the input data as microbial densities was found to facilitate the modeling procedure for the system monitoring even with a relatively lower number of input. Neural networks independently trained for each treatment tank and their subsequent combined data analysis allowed a successful prediction of the treatment system for at least two days.     

Production of xylitol from raw wood hydrolysates by Debaryomyces hansenii NRRL Y-7426

Xylose solutions were obtained from the hemicellulosic fraction of Eucalyptus globulus by means of sulphuric acid treatments performed under selected operational conditions. The hydrolysates were neutralized, supplemented with nutrients, sterilized and utilised as fermentation media for xylitol production using the yeast Debaryomyces hansenii NRRL Y-7426. Preliminary experiments allowed the adaptation of the strain to the culture media. Further fermentation trials were performed with adapted cells according to an incomplete, factorial design of experiments. Empirical models describing the bioconversion were derived from the experimental results. The effects of three operational variables (nutrient concentration, initial pH and shaking speed) on both xylose consumption and xylitol production rates were described by significant mathematical equations. Additional aspects in relation to the process were also discussed. The authors are grateful to Xunta de Galicia and “Orember” for their financial support to this work; as well as to Ms. Rocío Rodríguez Fontán for her excellent technical assistance.     

Bioprocess model building using artificial neural networks

Artificial neural networks are made upon of highly interconnected layers of simple neuron-like nodes. The neurons act as non-linear processing elements within the network. An attractive property of artificial neural networks is that given the appropriate network topology, they are capable of learning and characterising non-linear functional relationships. Furthermore, the structure of the resulting neural network based process model may be considered generic, in the sense that little prior process knowledge is required in its determination. The methodology therefore provides a cost efficient and reliable process modelling technique. One area where such a technique could be useful is biotechnological systems. Here, for example, the use of a process model within an estimation scheme has long been considered an effective means of overcoming inherent on-line measurement problems. However, the development of an accurate process model is extremely time consuming and often results in a model of limited applicability. Artificial neural networks could therefore prove to be a useful model building tool when striving to improve bioprocess operability. Two large scale industrial fermentation systems have been considered as test cases; a fed-batch penicillin fermentation and a continuous mycelial fermentation. Both systems serve to demonstrate the utility, flexibility and potential of the artificial neural network approach to process modelling.