Xylooligosaccharides: A Bibliometric Analysis and Current Advances of This Bioactive Food Chemical as a Potential Product in Biorefineries' Portfolios.

Xylooligosaccharides (XOS) are nondigestible compounds of great interest for food and pharmaceutical industries due to their beneficial prebiotic, antibacterial, antioxidant, and antitumor properties. The market size of XOS is increasing significantly, which makes its production from lignocellulosic biomass an interesting approach to the valorization of the hemicellulose fraction of biomass, which is currently underused. This review comprehensively discusses XOS production from lignocellulosic biomass, aiming at its application in integrated biorefineries. A bibliometric analysis is carried out highlighting the main players in the field. XOS production yields after different biomass pretreatment methods are critically discussed using Microsoft PowerBI® (2.92.706.0) software, which involves screening important trends for decision-making. Enzymatic hydrolysis and the major XOS purification strategies are also explored. Finally, the integration of XOS production into biorefineries, with special attention to economic and environmental aspects, is assessed, providing important information for the implementation of biorefineries containing XOS in their portfolio.

High stabilization and hyperactivation of a Recombinant ß-Xylosidase through Immobilization Strategies.

Attainment of a stable and highly active ß-xylosidase is of major importance for the efficient and cost-competitive hydrolysis of hemicellulose xylan, as well as for its industrial conversion into biofuels and biochemicals. Here, a recombinant ß-xylosidase of the glycoside hydrolase family (GH43) from Bacillus subtilis was produced in Escherichia coli culture, purified, and subsequently immobilized on agarose and chitosan. Glutaraldehyde and glyoxyl groups were evaluated as activating agents to select the most efficient derivative. Multi-point immobilization on agarose led to an extraordinary thermal stability (half-lives 3604 and 164-fold higher than the free enzyme, at 50° and 35 °C, respectively). Even for chitosan activated with glutaraldehyde, a low-cost support, thermal stability of the immobilized enzyme was 326 and 12-fold higher than the free enzyme at 50° and 35°C, respectively. Immobilized enzymes showed no release of any subunit for the agarose-glyoxyl derivative, and only a few ones for the support activated with glutaraldehyde. Most remarkably, the enzyme kinetic behavior after immobilization increased up to 4-fold in relation to the free one. ß-xylosidase, a tetrameric enzyme with four identical subunits, exists in equilibrium between the monomeric and oligomeric forms in solution. Depending on the pH of immobilization, the enzyme oligomerization can be favored, thus explaining the hyperactivation phenomenon. Both glyoxyl-agarose and chitosan-glutaraldehyde derivatives were used to catalyze corncob xylan hydrolysis, reaching 72 % conversion, representing a xylose productivity of around 20 g L-1 h-1. After ten 4h-cycles (pH 6.0, 35 °C), the xylan-to-xylose conversion remained approximately unchanged. Therefore, the immobilized ß-xylosidases prepared in this work can be of great interest as biocatalysts in a biorefinery context.

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

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.

High stabilization and hyperactivation of a recombinant β-Xylosidase through immobilization strategies

Attainment of a stable and highly active β-xylosidase is of major importance for the efficient and cost-competitive hydrolysis of hemicellulose xylan, as well as for its industrial conversion into biofuels and biochemicals. Here, a recombinant β-xylosidase of the glycoside hydrolase family (GH43) from Bacillus subtilis was produced in Escherichia coli culture, purified, and subsequently immobilized on agarose and chitosan. Glutaraldehyde and glyoxyl groups were evaluated as activating agents to select the most efficient derivative. Multi-point immobilization on agarose led to an extraordinary thermal stability (half-lives 3604 and 164-fold higher than the free enzyme, at 50° and 35 °C, respectively). Even for chitosan activated with glutaraldehyde, a low-cost support, thermal stability of the immobilized enzyme was 326 and 12-fold higher than the free enzyme at 50° and 35°C, respectively. Immobilized enzymes showed no release of any subunit for the agarose-glyoxyl derivative, and only a few ones for the support activated with glutaraldehyde. Most remarkably, the enzyme kinetic behavior after immobilization increased up to 4-fold in relation to the free one. β-xylosidase, a tetrameric enzyme with four identical subunits, exists in equilibrium between the monomeric and oligomeric forms in solution. Depending on the pH of immobilization, the enzyme oligomerization can be favored, thus explaining the hyperactivation phenomenon. Both glyoxyl-agarose and chitosan-glutaraldehyde derivatives were used to catalyze corncob xylan hydrolysis, reaching 72 % conversion, representing a xylose productivity of around 20 g L−1 h−1. After ten 4h-cycles (pH 6.0, 35 °C), the xylan-to-xylose conversion remained approximately unchanged. Therefore, the immobilized β-xylosidases prepared in this work can be of great interest as biocatalysts in a biorefinery context.

Repeated batches as a strategy for high 2G ethanol production from undetoxified hemicellulose hydrolysate using immobilized cells of recombinant Saccharomyces cerevisiae in a fixed-bed reactor.

BACKGROUND: The search for sustainable energy sources has become a worldwide issue, making the development of efficient biofuel production processes a priority. Immobilization of second-generation (2G) xylose-fermenting Saccharomyces cerevisiae strains is a promising approach to achieve economic viability of 2G bioethanol production from undetoxified hydrolysates through operation at high cell load and mitigation of inhibitor toxicity. In addition, the use of a fixed-bed reactor can contribute to establish an efficient process because of its distinct advantages, such as high conversion rate per weight of biocatalyst and reuse of biocatalyst. RESULTS: This work assessed the influence of alginate entrapment on the tolerance of recombinant S. cerevisiae to acetic acid. Encapsulated GSE16-T18SI.1 (T18) yeast showed an outstanding performance in repeated batch fermentations with cell recycling in YPX medium supplemented with 8 g/L acetic acid (pH 5.2), achieving 10 cycles without significant loss of productivity. In the fixed-bed bioreactor, a high xylose fermentation rate with ethanol yield and productivity values of 0.38 gethanol/gsugars and 5.7 g/L/h, respectively were achieved in fermentations using undetoxified sugarcane bagasse hemicellulose hydrolysate, with and without medium recirculation. CONCLUSIONS: The performance of recombinant strains developed for 2G ethanol production can be boosted strongly by cell immobilization in alginate gels. Yeast encapsulation allows conducting fermentations in repeated batch mode in fixed-bed bioreactors with high xylose assimilation rate and high ethanol productivity using undetoxified hemicellulose hydrolysate.

Performance targets defined by retro-techno-economic analysis for the use of soybean protein as saccharification additive in an integrated biorefinery.

The use of additives in the enzymatic saccharification of lignocellulosic biomass can have positive effects, decreasing the unproductive adsorption of cellulases on lignin and reducing the loss of enzyme activity. Soybean protein stands out as a potential lignin-blocking additive, but the economic impact of its use has not previously been investigated. Here, a systematic evaluation was performed of the process conditions, together with a techno-economic analysis, for the use of soybean protein in the saccharification of hydrothermally pretreated sugarcane bagasse in the context of an integrated 1G-2G ethanol biorefinery. Statistical experimental design methodology was firstly applied as a tool to select the process variable solids loading at 15% (w/w) and soybean protein concentration at 12% (w/w), followed by determination of enzyme dosage at 10 FPU/g and hydrolysis time of 24 h. The saccharification of sugarcane bagasse under these conditions enabled an increase of 26% in the amount of glucose released, compared to the control without additive. The retro-techno-economic analysis (RTEA) technique showed that to make the biorefinery economically feasible, some performance targets should be reached experimentally such as increasing biomass conversion to ideally 80% and reducing enzyme loading to 5.6 FPU/g in the presence of low-cost soybean protein.

Alternative Low-Cost Additives to Improve the Saccharification of Lignocellulosic Biomass.

A potential strategy to mitigate problems related to unproductive adsorption of enzymes onto lignin during the saccharification of lignocellulosic biomass is the addition of lignin-blocking agents to the hydrolysis reaction medium. However, there is a clear need to find more cost-effective additives for use in large-scale processes. Here, selected alternative low-cost additives were evaluated in the saccharification of steam-exploded sugarcane bagasse using a commercial enzymatic cocktail. The addition of soybean protein, tryptone, peptone, and maize zein had positive effects on glucose release during the hydrolysis, with gains of up to 36% when 8% (w/w) soybean protein was used. These improvements were superior to those obtained using bovine serum albumin (BSA), a much more expensive protein that has been widely reported for such an application. Moreover, addition of soybean protein led to a saving of 48 h in the hydrolysis, corresponding to a 66% decrease in the reactor operation time required. In order to achieve the same hydrolysis yield without the soybean additive, the enzyme loading would need to be increased by 50%. FTIR spectroscopy and nitrogen elemental analysis revealed that the additives probably acted to reduce unproductive binding of cellulolytic enzymes onto the lignin portion of the sugarcane bagasse.

Mapping Salmonella typhimurium pathways using 13C metabolic flux analysis.

In the last years, Salmonella has been extensively studied not only due to its importance as a pathogen, but also as a host to produce pharmaceutical compounds. However, the full exploitation of Salmonella as a platform for bioproduct delivery has been hampered by the lack of information about its metabolism. Genome-scale metabolic models can be valuable tools to delineate metabolic engineering strategies as long as they closely represent the actual metabolism of the target organism. In the present study, a 13C-MFA approach was applied to map the fluxes at the central carbon pathways of S. typhimurium LT2 growing at glucose-limited chemostat cultures. The experiments were carried out in a 2L bioreactor, using defined medium enriched with 20% 13C-labeled glucose. Metabolic flux distributions in central carbon pathways of S. typhimurium LT2 were estimated using OpenFLUX2 based on the labeling pattern of biomass protein hydrolysates together with biomass composition. The results suggested that pentose phosphate is used to catabolize glucose, with minor fluxes through glycolysis. In silico simulations, using Optflux and pFBA as simulation method, allowed to study the performance of the genome-scale metabolic model. In general, the accuracy of in silico simulations was improved by the superimposition of estimated intracellular fluxes to the existing genome-scale metabolic model, showing a better fitting to the experimental extracellular fluxes, whereas the intracellular fluxes of pentose phosphate and anaplerotic reactions were poorly described.

Defining research & development process targets through retro-techno-economic analysis: The sugarcane biorefinery case.

A new approach is reported for techno-economic analysis of lignocellulosic ethanol production. With this methodology, general targets for key process variables can be draw, a valuable feedback for Research & Development teams. An integrated first- and second-generation ethanol from sugarcane biorefinery is presented as a case study for the methodology, with the biomass pretreated by liquid hot water, followed by enzymatic hydrolysis of the cellulose fraction. The hemicellulose fraction may be either fermented or biodigested. The methodology was able to identify the main variables that affect the process global economic performance: enzyme load in the cellulose hydrolysis reactor, cellulose-to-glucose, and xylose-to-ethanol yields. Windows of feasible operation are the graphical output of the methodology, outlining regions to be further explored experimentally. One example of quantitative result is that the maximum feasible enzyme load was 11.3 FPU/gcellulose when xylose is fermented to ethanol and 7.7 FPU/gcellulose when xylose is biodigested.

Salmonella typhimurium and Escherichia coli dissimilarity: Closely related bacteria with distinct metabolic profiles.

Live attenuated strains of Salmonella typhimurium have been extensively investigated as vaccines for a number of infectious diseases. However, there is still little information available concerning aspects of their metabolism. S. typhimurium and Escherichia coli show a high degree of similarity in terms of their genome contents and metabolic networks. However, this work presents experimental evidence showing that significant differences exist in their abilities to direct carbon fluxes to biomass and energy production. It is important to study the metabolism of Salmonella to elucidate the formation of acetate and other metabolites involved in optimizing the production of biomass, essential for the development of recombinant vaccines. The metabolism of Salmonella under aerobic conditions was assessed using continuous cultures performed at dilution rates ranging from 0.1 to 0.67 h(-1), with glucose as main substrate. Acetate assimilation and glucose metabolism under anaerobic conditions were also investigated using batch cultures. Chemostat cultivations showed deviation of carbon towards acetate formation, starting at dilution rates above 0.1 h(-1). This differed from previous findings for E. coli, where acetate accumulation was only detected at dilution rates exceeding 0.4 h(-1), and was due to the lower rate of acetate assimilation by S. typhimurium under aerobic conditions. Under anaerobic conditions, both microorganisms mainly produced ethanol, acetate, and formate. A genome-scale metabolic model, reconstructed for Salmonella based on an E. coli model, provided a poor description of the mixed fermentation pattern observed during Salmonella cultures, reinforcing the different patterns of carbon utilization exhibited by these closely related bacteria.

Bioelectricity versus bioethanol from sugarcane bagasse: is it worth being flexible?

BACKGROUND: Sugarcane is the most efficient crop for production of (1G) ethanol. Additionally, sugarcane bagasse can be used to produce (2G) ethanol. However, the manufacture of 2G ethanol in large scale is not a consolidated process yet. Thus, a detailed economic analysis, based on consistent simulations of the process, is worthwhile. Moreover, both ethanol and electric energy markets have been extremely volatile in Brazil, which suggests that a flexible biorefinery, able to switch between 2G ethanol and electric energy production, could be an option to absorb fluctuations in relative prices. Simulations of three cases were run using the software EMSO: production of 1G ethanol + electric energy, of 1G + 2G ethanol and a flexible biorefinery. Bagasse for 2G ethanol was pretreated with a weak acid solution, followed by enzymatic hydrolysis, while 50% of sugarcane trash (mostly leaves) was used as surplus fuel. RESULTS: With maximum diversion of bagasse to 2G ethanol (74% of the total), an increase of 25.8% in ethanol production (reaching 115.2 L/tonne of sugarcane) was achieved. An increase of 21.1% in the current ethanol price would be enough to make all three biorefineries economically viable (11.5% for the 1G + 2G dedicated biorefinery). For 2012 prices, the flexible biorefinery presented a lower Internal Rate of Return (IRR) than the 1G + 2G dedicated biorefinery. The impact of electric energy prices (auction and spot market) and of enzyme costs on the IRR was not as significant as it would be expected. CONCLUSIONS: For current market prices in Brazil, not even production of 1G bioethanol is economically feasible. However, the 1G + 2G dedicated biorefinery is closer to feasibility than the conventional 1G + electric energy industrial plant. Besides, the IRR of the 1G + 2G biorefinery is more sensitive with respect to the price of ethanol, and an increase of 11.5% in this value would be enough to achieve feasibility. The ability of the flexible biorefinery to take advantage of seasonal fluctuations does not make up for its higher investment cost, in the present scenario.

Non-conventional induction strategies for production of subunit swine erysipelas vaccine antigen in rE. coli fed-batch cultures.

In spite of the large number of reports on fed-batch cultivation of E. coli, alternative cultivation/induction strategies remain to be more deeply exploited. Among these strategies, it could be mentioned the use of complex media with combination of different carbon sources, novel induction procedures and feed flow rate control matching the actual cell growth rate. Here, four different carbon source combinations (glucose, glycerol, glucose + glycerol and auto-induction) in batch media formulation were compared. A balanced combination of glucose and glycerol in a complex medium formulation led to: fast growth in the batch-phase; reduced plasmid instability by preventing early expression leakage; and protein volumetric productivity of 0.40 g.L(-1).h(-1). Alternative induction strategies were also investigated. A mixture of lactose and glycerol as supplementary medium fully induced a high biomass population, reaching a good balance between specific protein production (0.148 gprot.gDCW (-1)) and volumetric productivity (0.32 g.L(-1).h(-1)). The auto-induction protocol showed excellent results on specific protein production (0.158 gprot.gDCW (-1)) in simple batch cultivations. An automated feed control based on the on-line estimated growth rate was implemented, which allowed cells to grow at higher rates than those generally used to avoid metabolic overflow, without leading to acetate accumulation. Some of the protocols described here may provide a useful alternative to standard cultivation and recombinant protein production processes, depending on the performance index that is expected to be optimized. The protocols using glycerol as carbon source and induction by lactose feeding, or glycerol plus glucose in batch medium and induction by lactose pulse led to rSpaA production in the range of 6 g.L(-1), in short fed-batch processes (16 to 20 h) with low accumulation of undesired side metabolites.

Evaluation of immobilized lipases on poly-hydroxybutyrate beads to catalyze biodiesel synthesis.

Five microbial lipase preparations from several sources were immobilized by hydrophobic adsorption on small or large poly-hydroxybutyrate (PHB) beads and the effect of the support particle size on the biocatalyst activity was assessed in the hydrolysis of olive oil, esterification of butyric acid with butanol and transesterification of babassu oil (Orbignya sp.) with ethanol. The catalytic activity of the immobilized lipases in both olive oil hydrolysis and biodiesel synthesis was influenced by the particle size of PHB and lipase source. In the esterification reaction such influence was not observed. Geobacillus thermocatenulatus lipase (BTL2) was considered to be inadequate to catalyze biodiesel synthesis, but displayed high esterification activity. Butyl butyrate synthesis catalyzed by BTL2 immobilized on small PHB beads gave the highest yield (≈90 mmol L(-1)). In biodiesel synthesis, the catalytic activity of the immobilized lipases was significantly increased in comparison to the free lipases. Full conversion of babassu oil into ethyl esters was achieved at 72 h in the presence of Pseudozyma antarctica type B (CALB), Thermomyces lanuginosus lipase (Lipex(®) 100 L) immobilized on either small or large PHB beads and Pseudomonas fluorescens (PFL) immobilized on large PHB beads. The latter preparation presented the highest productivity (40.9 mg of ethyl esters mg(-1) immobilized protein h(-1)).

Monitoring bioreactors using principal component analysis: production of penicillin G acylase as a case study.

The complexity of biological processes often makes impractical the development of detailed, structured phenomenological models of the cultivation of microorganisms in bioreactors. In this context, data pre-treatment techniques are useful for bioprocess control and fault detection. Among them, principal component analysis (PCA) plays an important role. This work presents a case study of the application of this technique during real experiments, where the enzyme penicillin G acylase (PGA) was produced by Bacillus megaterium ATCC 14945. PGA hydrolyzes penicillin G to yield 6-aminopenicilanic acid (6-APA) and phenyl acetic acid. 6-APA is used to produce semi-synthetic beta-lactam antibiotics. A static PCA algorithm was implemented for on-line detection of deviations from the desired process behavior. The experiments were carried out in a 2-L bioreactor. Hotteling's T(2) was the discrimination criterion employed in this multivariable problem and the method showed a high sensibility for fault detection in all real cases that were studied.

Improving the performance of a continuous process for the production of ethanol from starch.

In a previous work, a continuous simultaneous saccharification and fermentation process to produce ethanol from cassava starch was studied, using a set of fixed-bed reactors. The biocatalyst consisted of glucoamylase immobilized in silica particles and co-immobilized with S. cerevisiae in pectin gel. Using 3.8 U mL(-1) (reactor) and 0.05 g(wet yeast) mL(-1) (reactor) at start-up, starch hydrolysis was the rate-limiting step. Maximum ethanol productivity was 5.8 g(ethanol) L(-1) h(-1), with 94.0% conversion of total reducing sugars (TRS) and 83.0% of the ethanol theoretical yield. In this work, the molar mass of the substrate and the biocatalyst particle size were reduced in an attempt to improve the bioreactor performance. The diameters of silica and pectin gel particles were reduced from 100 mum and 3-4 mm, respectively, to 60 mum and 1-1.5 mm, and the degree of substrate prehydrolysis by alpha-amylase was increased. The bioreactor performance was assessed for different loads of immobilized glucoamylase (2.1, 2.8, and 3.8 U mL(-1) (reactor)), for the same initial cell concentration (0.05 g(wet yeast.)mL(-1) (reactor)). Feeding with 154.0 g L(-1) of TRS and using 3.8 U mL(-1) (reactor), fermentation became the rate-limiting step. Productivity reached 11.7 g L(-1) h(-1), with 97.0% of TRS conversion and 92.0% of the ethanol theoretical yield. The reactor was operated during 275 h without any indication of destabilization.

Remote engineering for a cheese whey biorefinery: an Internet-based application for process design, economic analysis, monitoring, and control of multiple plant sites.

The proteolysis of cheese whey with the aid of immobilized enzymes is an attractive alternative for this by-product of the dairy industry. Among some possible applications for whey protein hydrolysates, one may cite their use as protein source for individuals with reduced capacity of digestion, or with genetic metabolic disorders (phenylketonuria patients, for instance). The multipurpose plant that processes whey is named here as a cheese whey biorefinery. This work presents the remote control and monitoring of the whey biorefineries using the Internet. In an integrated environment, the web application also enables simulation and economic analyses of the process. This technology might allow small companies to access a remote "engineering centre", with know-how on plant design and advanced control techniques. The idea can also be extended to large dairy companies, providing the remote control of geographically spread sites of production.

Immobilization of trypsin on chitosan gels: use of different activation protocols and comparison with other supports.

Trypsin was immobilized on chitosan gels coagulated with 0.1 or 1 M NaOH and activated with glutaraldehyde or glycidol. The derivatives were characterized by their recovered activity, thermal (40, 55 and 70 degrees C) and alkaline (pH 11) stabilities, amount of enzyme immobilized on gels for several enzyme loads (8-14 mg(protein)/g(Gel)) and compared to agarose derivatives. Enzyme loads higher than 14 mg(protein)/g(Gel) can be immobilized on glutaraldehyde derivatives, which showed 100% immobilization yield and, for loads up to 8 mg(protein)/g(Gel), 100% recovered activity. Activation with glycidol led to lower immobilization yields than the ones obtained with glutaraldehyde, 61% for agarose-glyoxyl (AgGly) with low grade of activation and 16% for the chitosan-glyoxyl (ChGly), but allowed obtaining the most stable derivative (ChGly), that was 660-fold more stable than the soluble enzyme at 55 and 70 degrees C-approximately threefold more stable than AgGly. The ChGly derivative presented also the highest stability during incubation at pH 11. Analyses of lysine residue contents in soluble and immobilized trypsin indicated formation of multipoint bonds between enzyme and support, for glyoxyl derivatives.

Online filtering of CO2 signals from a bioreactor gas outflow using a committee of constructive neural networks.

This work proposes a committee of Cascade Correlation neural networks as an online smoother for random measurement noise. The goals of this paper are twofold: first it intends to explore the possibilities of using a constructive neural network algorithm to learn how to filter typical noisy data from a bioreactor, CO(2) mol fractions of the effluent gas during the aerobic cultivation of Bacillus megaterium to produce the enzyme penicillin G acylase. Second, to propose a committee of such networks for achieving more realistic results, capturing the inherent trend of the process. In order to do that this paper discusses the advantages of using a constructive neural network algorithm, describes how the committee of NNs operates and evaluates its performance using real CO(2) online data obtained in laboratorial experiments. The paper also presents results obtained with classical filtering algorithms, for comparison.

Neural network inference of molar mass distributions of peptides during tailor-made enzymatic hydrolysis of cheese whey: effects of pH and temperature.

The fine-tuning of the enzymatic hydrolysis of proteins may provide a pool of peptides with predefined molar mass distributions. However, the complex mixture of molecules (peptides and amino acids) that results after the proteolysis of cheese whey turns unfeasible the assessment of individual species. In this work, a hybrid kinetic model for the proteolysis of whey by alcalase, multipoint-immobilized on agarose, is presented, which takes into account the influence of pH (8.0-10.4) and temperature (40-55 degrees C) on the activity of the enzyme. Five ranges of peptides' molar mass have their reaction rates predicted by neural networks (NNs). The output of NNs trained for constant pH and temperatures was interpolated, instead of including these variables in the input vector of a larger NN. Thus, the model complexity was reduced. Coupled to differential mass balances, this hybrid model can be employed for the online inference of peptides' molar mass distributions. Experimental kinetic assays were carried out using a pH-stat, in a laboratory-scale (0.03 L) batch reactor. The neural-kinetic model was integrated to a supervisory system of a bench-scale continually stirred tank reactor (0.5 L), providing accurate predictions during validation tests.

Comparing the performance of multilayer perceptrons networks and neuro-fuzzy systems for on-line inference of Bacillus megaterium cellular concentrations.

Penicillin G acylase (PGA) is one of the most important enzymes for the pharmaceutical industry. Bacillus megaterium has the advantage of producing extra-cellular PGA. This work compares two neural networks (NNs) architectures for on-line inference of B. megaterium cell mass in an aerated stirred tank bioreactor, during the production of PGA. Nowadays, intelligent computing tools such as artificial NNs and fuzzy logic are commonly applied for state inference and modeling of bioreactors. Combining these two approaches in hybrid, neuro-fuzzy systems, may be advantageous. Our results indicate that a neuro-fuzzy inference system showed a better performance to infer cell concentrations, when compared to multilayer perceptrons networks.