A transfer learning approach for predictive modeling of bioprocesses using small data.

Predictive modeling of new biochemical systems with small data is a great challenge. To fill this gap, transfer learning, a subdomain of machine learning that serves to transfer knowledge from a generalized model to a more domain-specific model, provides a promising solution. While transfer learning has been used in natural language processing, image analysis, and chemical engineering fault detection, its application within biochemical engineering has not been systematically explored. In this study, we demonstrated the benefits of transfer learning when applied to predict dynamic behaviors of new biochemical processes. Two different case studies were presented to investigate the accuracy, reliability, and advantage of this innovative modeling approach. We thoroughly discussed the different transfer learning strategies and the effects of topology on transfer learning, comparing the performance of the transfer learning models against benchmark kinetic and data-driven models. Furthermore, strong connections between the underlying process mechanism and the transfer learning model's optimal structure were highlighted, suggesting the interpretability of transfer learning to enable more accurate prediction than a naive data-driven modeling approach. Therefore, this study shows a novel approach to effectively combining data from different resources for bioprocess simulation.

Kinetic and hybrid modeling for yeast astaxanthin production under uncertainty.

Astaxanthin is a high-value compound commercially synthesized through Xanthophyllomyces dendrorhous fermentation. Using mixed sugars decomposed from biowastes for yeast fermentation provides a promising option to improve process sustainability. However, little effort has been made to investigate the effects of multiple sugars on X. dendrorhous biomass growth and astaxanthin production. Furthermore, the construction of a high-fidelity model is challenging due to the system's variability, also known as batch-to-batch variation. Two innovations are proposed in this study to address these challenges. First, a kinetic model was developed to compare process kinetics between the single sugar (glucose) based and the mixed sugar (glucose and sucrose) based fermentation methods. Then, the kinetic model parameters were modeled themselves as Gaussian processes, a probabilistic machine learning technique, to improve the accuracy and robustness of model predictions. We conclude that although the presence of sucrose does not affect the biomass growth kinetics, it introduces a competitive inhibitory mechanism that enhances astaxanthin accumulation by inducing adverse environmental conditions such as osmotic gradients. Moreover, the hybrid model was able to greatly reduce model simulation error and was particularly robust to uncertainty propagation. This study suggests the advantage of mixed sugar-based fermentation and provides a novel approach for bioprocess dynamic modeling.

Review of advanced physical and data-driven models for dynamic bioprocess simulation: Case study of algae-bacteria consortium wastewater treatment.

Microorganism production and remediation processes are of critical importance to the next generation of sustainable industries. Undertaking mathematical treatment of dynamic biosystems operating at any spatial or temporal scale is essential to guarantee their performance and safety. However, constructing physical models remains a challenge due to the extreme complexity of process biological mechanisms. Data-driven models also encounter severe limitations because datasets from large-scale bioprocesses are often scarce without complete information and on a restricted operational space. To fill this gap, the current research compares the performance of advanced physical and data-driven models for dynamic bioprocess simulations subject to incomplete and scarce datasets, which to the best of our knowledge has never been addressed before. In specific, kinetic models were constructed by integrating different classic models, and state-of-the-art hyperparameter selection frameworks were developed to design artificial neural networks and Gaussian process regression models. An algae-bacteria consortium wastewater treatment process was selected to test the accuracy of these modeling strategies, as it is one of the most sophisticated biosystems due to the intricate mutualistic and competitive interactions. Based on the current results and available data, a heuristic model selection procedure is provided. This study paves the way to facilitate future bioprocess modeling.

Comparison of physics-based and data-driven modelling techniques for dynamic optimisation of fed-batch bioprocesses.

The development of digital bioprocessing technologies is critical to operate modern industrial bioprocesses. This study conducted the first investigation on the efficiency of using physics-based and data-driven models for the dynamic optimisation of long-term bioprocess. More specifically, this study exploits a predictive kinetic model and a cutting-edge data-driven model to compute open-loop optimisation strategies for the production of microalgal lutein during a fed-batch operation. Light intensity and nitrate inflow rate are used as control variables given their key impact on biomass growth and lutein synthesis. By employing different optimisation algorithms, several optimal control sequences were computed. Due to the distinct model construction principles and sophisticated process mechanisms, the physics-based and the data-driven models yielded contradictory optimisation strategies. The experimental verification confirms that the data-driven model predicted a closer result to the experiments than the physics-based model. Both models succeeded in improving lutein intracellular content by over 40% compared to the highest previous record; however, the data-driven model outperformed the kinetic model when optimising total lutein production and achieved an increase of 40-50%. This indicates the possible advantages of using data-driven modelling for optimisation and prediction of complex dynamic bioprocesses, and its potential in industrial bio-manufacturing systems.

Overproduction of L-tryptophan via simultaneous feed of glucose and anthranilic acid from recombinant Escherichia coli W3110: Kinetic modeling and process scale-up.

L-tryptophan is an essential amino acid widely used in food and pharmaceutical industries. However, its production via Escherichia coli fermentation suffers severely from both low glucose conversion efficiency and acetic acid inhibition, and to date effective process control methods have rarely been explored to facilitate its industrial scale production. To resolve these challenges, in the current research an engineered strain of E. coli was used to overproduce L-tryptophan. To achieve this, a novel dynamic control strategy which incorporates an optimized anthranilic acid feeding into a dissolved oxygen-stat (DO-stat) glucose feeding framework was proposed for the first time. Three original contributions were observed. Firstly, compared to previous DO control methods, the current strategy was able to inhibit completely the production of acetic acid, and its glucose to L-tryptophan yield reached 0.211 g/g, 62.3% higher than the previously reported. Secondly, a rigorous kinetic model was constructed to simulate the underlying biochemical process and identify the effect of anthranilic acid on both glucose conversion and L-tryptophan synthesis. Finally, a thorough investigation was conducted to testify the capability of both the kinetic model and the novel control strategy for process scale-up. It was found that the model possesses great predictive power, and the presented strategy achieved the highest glucose to L-tryptophan yield (0.224 g/g) ever reported in large scale processes, which approaches the theoretical maximum yield of 0.227 g/g. This research, therefore, paves the way to significantly enhance the profitability of the investigated bioprocess.

An efficient model construction strategy to simulate microalgal lutein photo-production dynamic process.

Lutein is a high-value bioproduct synthesized by microalga Desmodesmus sp. It has great potential for the food, cosmetics, and pharmaceutical industries. However, in order to enhance its productivity and to fulfil its ever-increasing global market demand, it is vital to construct accurate models capable of simulating the entire behavior of the complicated dynamics of the underlying biosystem. To this aim, in this study two highly robust artificial neural networks (ANNs) are designed for the first time. Contrary to conventional ANNs, these networks model the rate of change of the dynamic system, which makes them highly relevant in practice. Different strategies are incorporated into the current research to guarantee the accuracy of the constructed models, which include determining the optimal network structure through a hyper-parameter selection framework, generating significant amounts of artificial data sets by embedding random noise of appropriate size, and rescaling model inputs through standardization. Based on experimental verification, the high accuracy and great predictive power of the current models for long-term dynamic bioprocess simulation in both real-time and offline frameworks are thoroughly demonstrated. This research, therefore, paves the way to significantly facilitate the future investigation of lutein bioproduction process control and optimization. In addition, the model construction strategy developed in this research has great potential to be directly applied to other bioprocesses. Biotechnol. Bioeng. 2017;114: 2518-2527. © 2017 Wiley Periodicals, Inc.

Daptomycin antibiotic production processes in fed-batch fermentation by Streptomyces roseosporus NRRL11379 with precursor effect and medium optimization.

Sodium decanoate was first found to be an effective precursor for synthesis of daptomycin from Streptomyces roseosporus NRRL11379 which was increased to 71.55-fold, compared with decanoic acid. The optimal flow rate of precursor was at 600 mg/(L day) after 48 h fermentation. From protein analysis via SDS-PAGE and identification of Tandem MS/MS afterwards, it deciphered that guanosine pentaphosphate synthetase, PNPase, tripeptidylamino peptidase primarily dealing with daptomycin synthesis. By applying Taguchi's L16 in culture optimization, the best yield was obtained from the medium with 60 g/L dextrin, 10 g/L dextrose, 1.0 g/L molasses, and 8 g/L yeast extract, respectively. The fed-batch fermentation, applied with feedback control of dextrin, stimulated the production up to 812 mg/L at 288 h. To our best knowledge, the daptomycin production in this study is significantly higher than that in previous studies and can make it more widely used in pharmaceutical industry.

Exploring engineering reduced graphene oxide-titanium dioxide (RGO-TiO2) nanoparticles treatment to effectively enhance lutein biosynthesis with Chlorella sorokiniana F31 under different light intensity.

The Chlorella sorokiniana F31 is a promising lutein producer with high lutein content. Herein, different graphene/TiO2 nanoparticles (NPs) were designed and synthesized by hydrothermal method. Through the UV-vis diffuse reflectance spectra (DRS) analysis, the results showed that RGO-TiO2 NPs can effectively expand visible light absorption compared with TiO2 NPs. Subsequently, the effects of these NPs on light utilization and lutein accumulation of C. sorokiniana F31 were investigated, and the RGO-TiO2 NPs treatment exhibited the higher lutein production and content than that of TiO2 and control group. As the optimal RGO-TiO2 (0.5 wt%) NPs concentration of 50 mg/L and light intensity of 211 µmol/m2/s, the supreme lutein content (15.55 mg/g), production (77.2 mg/L) and productivity (12.87 mg/L/d) were achieved. The performances are higher than most of reported values in previous study, indicated that RGO-TiO2 (0.5 wt%) NPs treatment is a promised strategy to enhance microalgal growth and lutein accumulation.

Novel Silanized Graphene Oxide/TiO2 Multifunctional Nanocomposite Photocatalysts: Simultaneous Removal of Cd2+ and Photodegradation of Phenols under Visible Light Irradiation.

Reduced graphene oxide (RGO)-TiO2 nanocomposites have exhibited effective photocatalytic degradation of various organic pollutants. However, their poor solubility could limit their application in water and other organic solvents. In this study, new graphene-based cross-linked ethylenediaminetetraacetic acid (EDTA)-RGO-TiO2 (ERGT) nanocomposites were synthesized for the removal of Cd(II) and photodegradation of phenol from wastewater by surface-functionalized cross-linking heavy metal chelating agent sodium edetate (EDTA) and photocatalyst titanium dioxide. The structural properties of fabricated nanocomposites were characterized using SEM, TEM, XPS, FTIR, XRD, UV-vis, gas sorption, and Raman spectroscopy analyses. Moreover, the adsorption of Cd(II) and the degradation of phenol under different conditions were studied. The experimental results revealed that the optimal catalytic degradation and adsorption performance could be achieved at pH 5.5, and the maximum absorption ratio of cadmium ions and the degradation efficiency of phenol can reach 178.2 mg/g and 90%, respectively. The results suggested that ERGT is a potential material for the removal of threatening pollutants from wastewater.

Development of a fourth-order compact finite difference scheme for simulation of simulated-moving-bed process.

A fourth-order compact finite difference scheme was developed to solve the model equation of simulated moving bed, which has a boundary condition that is updated along the calculation process and cannot be described as an explicit function of time. Two different methods, direct method and pseudo grid point method, were proposed to deal with the boundary condition. The high accuracy of the two methods was confirmed by a case study of solving an advection-diffusion equation with exact solution. The developed compact finite difference scheme was then used to simulate the SMB processes for glucose-fructose separation and enantioseparation of 1,1'-bi-2-naphtol. It was found that the simulated results fit well with the experimental data. Furthermore, the developed method was further combined with the continuous prediction method to shorten the computational time and the results showed that, the computational time can be saved about 45%.

Exploring biostimulation of plant hormones and nitrate supplement to effectively enhance biomass growth and lutein production with thermo-tolerant Desmodesmus sp. F51.

In this study, the interactive effect of plant hormone-salicylic acid and succinic acid on biomass growth, lutein content, and productivity of Desmodesmus sp. F51 were investigated. The results demonstrated that the synergistic action of salicylic acid and succinic acid could effectively enhance the assimilation of nitrate and significantly improve lutein production. The maximal lutein content 7.01 mg/g and productivity 5.11 mg/L/d could be obtained with a supplement of 100 µM salicylic acid and 2.5 mM succinic acid in batch culture. Furthermore, operation strategy of nitrate fed-batch coupled with supplementation for succinic acid and salicylic acid resulted in further enhancement of lutein content and productivity by 7.50 mg/g and 5.78 mg/L/d, respectively. The performance is better than most of the previously reported values.

Enhancement of cell growth and phycocyanin production in Arthrospira (Spirulina) platensis by metabolic stress and nitrate fed-batch.

Arthrospira (Spirulina) platensis is known to have high-quality proteins content and phycocyanin as one of the major pigment constituents of the cells, and the most challenging problem associated with phycocyanin production in Arthrospira is to optimize its intracellular accumulation. The present study evaluated the metabolic stress conditions (by nutrient enrichment) of Arthrospira platensis FACHB-314 for boosting biomass growth and high content phycocyanin accumulation. Experimental results showed that 5 mM sodium glutamate and 7.5 mM succinic acid could enhance biomass yield as well as phycocyanin accumulation compared with that of the control groups. The present study demonstrates that the biomass growth and phycocyanin accumulation were significantly enhanced in fed-batch cultivation of Arthrospira platensis by applying the substrates as metabolic stress agents combined with nitrate feeding strategy. cobA/hemD, hemG and ho genes presented the over-expression level with adding sodium glutamate and succinic acid in cultures, respectively, compared to the control groups.

Biocatalytic synthesis of vitamin A palmitate using immobilized lipase produced by recombinant Pichia pastoris.

In this work, the Candida antarctica lipase B (CALB), produced by recombinant Pichia pastoris, was immobilized and used to synthesize vitamin A palmitate by transesterification of vitamin A acetate and palmitic acid in organic solvent. The reaction conditions including the type of solvent, temperature, rotation speed, particle size, and molar ratio between the two substrates were investigated. It turned out that the macroporous resin HPD826 serving as a carrier showed the highest activity (ca. 9200 U g-1) among all the screened carriers. It was found that the transesterification kinetic of the immobilized CALB followed the ping pong Bi-Bi mechanism and the reaction product acetic acid inhibited the enzymatic reaction with an inhibition factor of 2.823 mmol L-1. The conversion ability of the immobilized CALB was 54.3% after 15 cycles. In conclusion, the present work provides a green route for vitamin A palmitate production using immobilized CALB to catalyze the transesterification of vitamin A acetate and palmitic acid.

Fed-batch strategy for enhancing cell growth and C-phycocyanin production of Arthrospira (Spirulina) platensis under phototrophic cultivation.

The C-phycocyanin generated in blue-green algae Arthrospira platensis is gaining commercial interest due to its nutrition and healthcare value. In this study, the light intensity and initial biomass concentration were manipulated to improve cell growth and C-phycocyanin production of A.platensis in batch cultivation. The results show that low light intensity and high initial biomass concentration led to increased C-phycocyanin accumulation. The best C-phycocyanin productivity occurred when light intensity and initial biomass concentration were 300µmol/m(2)/s and 0.24g/L, respectively. The fed-batch cultivation proved to be an effective strategy to further enhance C-phycocyanin production of A.platensis. The results indicate that C-phycocyanin accumulation not only requires nitrogen-sufficient condition, but also needs other nutrients. The highest C-phycocyanin content (16.1%), production (1034mg/L) and productivity (94.8mg/L/d) were obtained when using fed-batch strategy with 5mM medium feeding.

Direct proteomic mapping of Streptomyces roseosporus NRRL 11379 with precursor and insights into daptomycin biosynthesis.

This first-attempt study provided liquid chromatography tandem mass (LC-MS/MS) proteomics approach to explore precursor effects on daptomycin synthesis from Streptomyces roseosporus NRRL 11379. Among all, 357 and 691 differential proteins from 601 proteins in precursor group (144 h+) and 935 proteins in non-precursor group (144 h-) were identified, respectively. Through the simulation of the 2D-protein mapping, most proteins were found in isoelectric points ranged of 4.5-10.0 as well as Mws ranged 10-100 kDa. As a result, LC-MS/MS analysis was consistence with the analytical results of two-dimensional electrophoresis (2DE) but provided much intact profiles of proteins by precursor effect on S. roseosporus. To have more insight exploration, differential proteins associated to Streptomyces spp. were defined into 14 groups of their functional classification. The major differential proteins were in transport/membrane functional group with an occupation of 12.4% for 144 h+ and 5.2% for 144 h-, respectively. LC-MS/MS results as a direct proteomic mapping approach reveal more daptomycin synthetic and regulation-related proteins from precursor group in terms of methyltransferase, ATP-binding cassette (ABC) transporters, resistance proteins and regulators.

Solubility properties and diffusional extraction behavior of natamycin from Streptomyces gilvosporeus biomass.

Natamycin is a type of polyene macrolide antibiotic and has been produced in submerged microbial cultures of some natural Streptomyces strains. Natamycin extraction from cellular biomass is greatly affected by the molecular and solubilization characteristics of the extraction solvent, and this is a major reason for the routine attainment of low volumetric titers, resulting from sparing natamycin solubility. In this work, a series of experiments were conducted to investigate the solubility of natamycin in some selected organic solvents in order to assess the influence on natamycin extraction yield. Natamycin showed the highest solubility in 75% aqueous methanol under the conditions of pH 2, 30°C and 1 atm. Furthermore, the extraction of natamycin using 75% aqueous methanol was performed and the highest extraction yield of 45.7% was obtained under pH 2. A mathematical model derived from Fick's law of the biomolecular diffusion process was developed to fit the experimental kinetic data of natamycin extraction.

Phototrophic cultivation of a thermo-tolerant Desmodesmus sp. for lutein production: effects of nitrate concentration, light intensity and fed-batch operation.

Four indigenous thermo-tolerant Desmodesmus sp. strains were examined for their ability to produce lutein. Among them, Desmodesmus sp. F51 was the best strain for this purpose. The medium composition, nitrate concentration and light intensity were manipulated to improve the phototrophic growth and lutein production of Desmodesmus sp. F51. It was found that a nitrogen-sufficient condition was required for lutein accumulation, while a high light intensity enhanced cell growth but caused a decrease in the lutein content. The best cell growth and lutein production occurred when the light intensity and initial nitrate concentration were 600 µmol/m(2)/s and 8.8 mM, respectively. The fed-batch cultivation strategy was shown to further improve lutein production. The highest lutein productivity (3.56±0.10 mg/L/d) and content (5.05±0.20 mg/g) were obtained when pulse-feeding of 2.2 mM nitrate was employed. This study demonstrated the potential of using Desmodesmus sp. F51 as a lutein producer in practical applications.

High-level expression of recombinant glucose dehydrogenase and its application in NADPH regeneration.

Two glucose dehydrogenase (E.C. 1.1.1.47) genes, gdh223 and gdh151, were cloned from Bacillus megaterium AS1.223 and AS1.151, and were inserted into pQE30 to construct the expression vectors, pQE30-gdh223 and pQE30-gdh151, respectively. The transformant Escherichia coli M15 with pQE30-gdh223 gave a much higher glucose dehydrogenase activity than that with the plasmid pQE30-gdh151. Thus it was used to optimize the expression of glucose dehydrogenase. An proximately tenfold increase in GDH activity was achieved by the optimization of culture and induction conditions, and the highest productivity of glucose dehydrogenase (58.7 U/ml) was attained. The recombinant glucose dehydrogenase produced by E. coli M15 (pQE30-gdh223) was then used to regenerate NADPH. NADPH was efficiently regenerated in vivo and in vitro when 0.1 M glucose was supplemented concomitantly in the reaction system. Finally, this coenzyme-regenerating system was coupled with a NADPH-dependent bioreduction for efficient synthesis of ethyl (R)-4-chloro-3-hydroxybutanoate from ethyl 4-chloro-3-oxobutanoate.

Construction of a two-strain system for asymmetric reduction of ethyl 4-chloro-3-oxobutanoate to (S)-4-chloro-3-hydroxybutanoate ethyl ester.

Escherichia coli M15 (pQE30-car0210) was constructed to express carbonyl reductase (CAR) by cloning the car gene from Candida magnoliae and inserting it into pQE30. By cultivating E. coli M15 (pQE30-car0210) and M15 (pQE30-gdh0310), 8.2-fold and 12.3-fold enhancements in specific enzymatic activity over the corresponding original strain were achieved, respectively. After separate cultivations, these two strains were then mixed together at appropriate ratio to construct a novel two-strain system, in which M15 (pQE30-car0210) expressed CAR for ethyl 4-chloro-3-oxobutanoate (COBE) bioreduction and M15 (pQE30-gdh0310) expressed glucose dehydrogenase (GDH) for nicotinamide adenine dinucleotide phosphate (NADPH) regeneration. In this complex system, the effects of substrate concentration, the biomass ratio between two strains as well as reaction temperature were investigated for efficient bioreduction. The results showed that the bioreduction reaction could be completed effectively without any addition of GDH or NADPH/NADP(+). An optical purity of 99% (enantiometric efficiency) was obtained, and the yield of (S)-4-chloro-3-hydroxybutanoate ethyl ester reached 96.6% when initial concentration of COBE was 36.9 mM. The coupling reactions between two different strains were further explored by determining the profile of NADPH in the reaction broth.

Asymmetric reduction of ethyl 4-chloro-3-oxobutanoate to ethyl (R)-4-chloro-3-hydroxybutanoate with two co-existing, recombinant Escherichia coli strains.

Two recombinant strains, E. coli M15 (pQE30-alr0307) and E. coli M15 (pQE30-gdh0310), which were constructed to express, respectively, an NADPH-dependent aldehyde reductase gene and a glucose dehydrogenase gene, were mixed in an appropriate ratio and used for the asymmetric reduction of ethyl 4-chloro-3-oxobutanoate to ethyl (R)-4-chloro-3-hydroxybutanoate. The former strain acted as catalyst and the latter functioned in NADPH regeneration. The biotransformation was completed effectively without any addition of glucose dehydrogenase or NADP+/NADPH. An optical purity of 99% (ee) was obtained and the product yield reached 90.5% from 28.5 mM: substrate.