Keywords: Modelling and Identification, Dynamics and Control, Integrated Bioprocessing
Abstract: The increasing environmental pollution, the deficite of natural ressources and global warming are just some of many challanges the modern world has to cope with. One key factor for overcoming these pressing societal challenges is the replacement of inefficent production processes by alternative and innovative concepts. Biotechnology can deliver such innovative approaches and products, as the employment of living organisms allows to develop processes which utilize renewable substances, are based on low energy concepts and produce a minimum of undesired pollutive and contaminative side-products. Despite significant methodical and experimental developments over the last decades, the structured analysis, design and optimization of biotechnological processes is still challenging. The involved intra- and extracellular interactions and the present regulatory networks are complex. Mathematical modeling, model based analysis, design, and optimization can provide means towards economic operation of biotechnological processes. However, while the always-present regulatory mechanisms allow biological organisms to adapt to changing environmental conditions and to cope with perturbations, they limit the excitability. This, together with the measurement uncertainties and the complexity render the development of mathematical models difficult. This dilemma motivates this perspective. After a brief summary of existing modeling and validation approaches we focus on so called set-based modeling and analysis methods. Specifically we outline the main ideas behind set-based modeling and model invalidation of biotechnological processes using infeasibility certificates. We furthermore present a top-down modeling approach for growth processes, which directly exploits the potential of set-based methods. The presented methods are underlined considering two practically relevant biotechnological processes. The work is finalized by a perspective on the potential of set-based approaches with respect to monitoring, prediction, analysis, and guaranteed control.

Keywords: Systems Biology, Metabolic Engineering, Modelling and Identification
Abstract: Macroscopic and metabolic models are ubiquitous tools for bioprocess modeling and understanding. While these two modeling approaches bring answers to different objectives, a remaining challenging question is how to couple macroscopic and intracellular models to enlarge the prediction capabilities of the model while keeping a model structure with low complexity level? In this paper, we provide conceptual guidelines to address this question. We propose a new modeling framework that manages non-balanced growth condition and hence accumulation of intracellular metabolites. The first stage consists in determining metabolic sub-networks, which are assumed to satisfy balanced growth condition. Then, the remaining metabolites interconnecting the sub-networks can dynamically accumulate or be reused. The basis of our approach is illustrated for the carbon metabolic network of unicellular microalgae.

Keywords: Microbial Technology
Abstract: The availability of parallel fermentation systems comprised of miniature, independently controlled bioreactors provides new opportunities for high throughput bioprocess development. In this study, we demonstrated the use of a four bioreactor system to validate predictions from a dynamic flux balance model of Saccharomyces cerevisiae metabolism. First we showed that the four 250 mL bioreactors generated very reproducible aerobic batch culture data and that the parallel system results could be accurately scaled-up to a standard 1.25 L laboratory bioreactor by matching oxygen mass transfer coefficients in the different reactors. A S. cerevisiae dynamic flux balance model previously developed in our group was shown to produce anaerobic and aerobic batch profiles in excellent agreement with the parallel system. The validated model was used to determine the optimal aerobic-anaerobic switching time for maximal ethanol production in batch culture. An optimal switching time in agreement with parallel system experiments was obtained. We concluded that parallel fermentation is a powerful tool for batch culture optimization when used in conjunction with dynamic metabolic models.

Keywords: Metabolic Engineering, Modelling and Identification, Food Engineering
Abstract: This paper presents the metabolic-fluxes calculation of a metabolic network representing the central metabolism of the yeast Saccharomyces cerevisiae in the wine-making context. Two solution methods are compared: a metabolic flux analysis (MFA) using convex analysis and providing narrow intervals of variation for the fluxes, and a flux balance analysis (FBA) based on an objective function. The constraints allowing the solution of the underdetermined set of algebraic equations are typically originating from measurements of uptake and secretion rates of external metabolites, and/or the use of an objective function, and/or metabolic constraints. It is shown here that converting reactions schemes in Cmol unit combined with data reconciliation provides a convenient formulation of closed carbon balance. In this form, the constraint formulation is natural and facilitates the understanding of the carbon distribution within the yeast metabolism.

Keywords: Systems Biology, Modelling and Identification, Metabolic Engineering
Abstract: The systems biology approach was considered for the analysis of metabolic regulation in Escherichia coli. In particular, it was shown that the feed-forward and feed-back regulations are formed in the metabolic regulation by enzymatic and transcriptional regulations, where these contribute to the robustness of the cell system. Moreover, the effects of the specific pathway gene knockout such as Δpgi, Δzwf, and Δpyk on the metabolic regulations were clarified based on systems biology approach. Mathematical modeling with computer simulation was also made for the main metabolism.

Keywords: Microbial Technology, Systems Biology, Metabolic Engineering
Abstract: Since the metabolic network reconstruction of Haemophilus influenzae was published in 1999, many other researchers have followed this trend. The possibilities and potential of having an in silico metabolic network for a specific organism is gaining emphasis among researchers that see the promise that the new era of genome-scale metabolic models could bring to the scientific scene. Enterococcus faecalis is a Gram positive bacterium from the lactic acid group that inhabits commensally the gut of humans and warm-blooded animals. However, it acquired a relevant position in the medical scenery as the main cause of nosocomial infections in hospital environments and the reason behind multiple diseases, namely urinary tract infections, endocarditis, meningitis, to name a few. In this work the first genome-scale model of E. faecalis, has been reconstructed to serve as a valuable tool to predict and better understand the phenotypic behaviour and its metabolism. The reconstruction process resulted in a network with 603 reactions and 576 metabolites. From the annotated genome it was possible to identify 366 genes with enzymatic activity. Performed simulations allowed us to validate the formation of end-products, as well as determine critical genes involved in the metabolism of this pathogen.

Keywords: Systems Biology
Abstract: Cells in the brain carry out massively parallel computation with minuscule energy consumption. The popular view of brain computation imagines that this, like computers, happens through electrical signals. In fact chemical information processing is if anything, even more powerful than electrical when we get down to subcellular computations. I will discuss the basis for electrical and chemical computation as they occur in the cell. I will then zoom in on how two key operations in memory are implemented in chemistry: pattern selection, and stable storage of activity states. I will show that as often happens in Biology, the chemical and electrical levels of computation are so tightly intertwined as to be the same, enormously efficient computational system. Finally I will discuss recent developments in imaging and physiology that have unleashed a flood of data to tell us about the circuitry and activity of the brain, leaving us with the colossal challenge of integrating all of this in computer models.

Keywords: Food Engineering, Dynamics and Control
Abstract: A major drawback of microwave processing is the heterogeneity of treatment, which prevents from a plenty benefit of its flexibility and rapidity. Most of time, this operation is realized in continuous processes, composed of a series of microwave generators with adjustable power. In this paper is proposed a methodology leading to an optimal setting of these powers in order to warrant the expected microorganisms’ inactivation during simultaneous cooking and pasteurization, while preserving quality. It consists in minimizing a multicriteria formulation including hottest and coldest points on the first hand, and final logarithmic inactivation on the other one. The simulation model is composed of a reduction of the heat equation via a finite volume scheme with a source term deduced from appropriate closed-form solutions of the Maxwell’s equations, whereas the non-isothermal inactivation is described by the Geeraerd model. The methodology is carried out by considering treatment of minced beef.

Keywords: Systems Biology, Modelling and Identification, Metabolic Engineering
Abstract: Abstract: The regulatory action of insulin on the blood glucose homeostasis is mediated by the Insulin signalling pathway. To quantify this regulation, we have integrated the models of Insulin secretion, insulin signalling, mTOR signalling and blood glucose uptake by various tissues. We have analyzed the effect of perturbations in the integrated insulin-mTOR signalling pathway on blood glucose levels. These effects were studied in the tissues in which glucose uptake is dependent on insulin (i.e. muscle, adipose and heart). Seventy five rate parameters in the integrated network were perturbed and the corresponding steady state plasma glucose levels were recorded. The fold changes in these parameters leading to pre-diabetic and diabetic states were characterized. Perturbations in 22 parameters elicited diabetic effect and 31 parameters showed pre-diabetic effects for certain fold change in basal parameter values. In the insulin signalling pathway, the concentrations and phosphorylation states of PTP and PIP3 were most effective in leading to diabetic state with only 3 and -3 fold change in their parameters, respectively. While increasing concentrations of PTEN by five folds also led to diabetic state, decreasing the phosphorylation of IRS, AKT and GSK3 unto 10 folds resulted in diabetic state. In the mTOR pathway, increasing the influence of amino acid on Rheb-GTP localization to the inhibitory complex by 4 folds elicited diabetic state. On the other hand decreasing conversion of RhebGTP to RhebGDP and formation of mTOR-raptor-PRAS40 complex also resulted in diabetic state. This demonstrates the influence of excess amino acid intake on diabetic state. Overall, it was noted that only 2 to 3 fold change in sensitive parameters resulted in pre-diabetic state.

Keywords: Environmental Processes (Wastewater, Bioremediation), Parameter and State Estimation
Abstract: Microalgae culture is used in various biotechnological applications. Optimisation of the system productivity needs reliable sensors. However, physical sensors for biomass and dissolved dioxide carbon concentrations are expensive and not accurate, especially for online measurements. In this context, robust and efficient software sensors have to be developed. In this paper, an Unscented Kalman filter (UKF) methodology is proposed to estimate components concentrations in a photobioreactor. The microalgae Chlamydomonas reinhardtii is used as model organism. The aim of this paper is to develop an online software estimator that reconstructs the biomass, carbon dioxide and oxygen concentrations in the liquid phase, from online measurements of components molar fraction in the output gas provided by a mass spectrometer. The proposed estimator is validated through experimental data collected on a lab-scale photobioreactor.

Keywords: Modelling and Identification, Bulk Chemicals Production
Abstract: This paper demonstrates the added benefits of using uncertainty and sensitivity analysis in the kinetics of enzymatic biodiesel production. For this study, a kinetic model by Fedosov and co-workers is used. For the uncertainty analysis the Monte Carlo procedure was used to statistically quantify the variability in the model outputs due to uncertainties in the parameter estimates; showing the model is most reliable in the start (first 5 hours) of the reaction. To understand which input parameters are responsible for the output uncertainty, two global sensitivity methods (Standardized Regression Coefficients, and Morris screening) were used. The results from both sensitivity analyses identified that only 10 of the 32 parameters are influential to the model outputs. The model was then simplified by removing the non-influential parameters. A parity plot of the simplified model vs. the full model gave a R2 value of over 0.95 for all the model outputs.

Keywords: Modelling and Identification
Abstract: A defined mixed culture of specialized microbes that exploits the native capabilities of each member species is a promising alternative to use of a single engineered microbe for cellulosic biofuels production. We explored such a synthetic consortium that couples the high cellulolytic activity of the filamentous fungus Trichoderma reesei with the ability of the yeasts Saccharomyces cerevisiae and Scheffersomyces stipitis to ferment hexose and pentose sugars to ethanol. Consortium stability was demonstrated by culturing the three microbes on a mixture of cellulose and xylan. As a first step towards understanding and manipulating this consortium, we developed a simple dynamic model with unstructured descriptions of enzyme synthesis, cellulose and hemicellulose degradation, sugar uptake, cell growth, and ethanol production. The batch culture model contained 10 ordinary differential equations with parameters obtained from the literature and experiment to the extent possible. The dynamic model was used to predict initial concentration of each cell type that maximized ethanol productivity for a fixed total inoculum concentration. The simulated ratio of cellulose to hemicellulose in the feedstock was varied to determine the effects on the optimal inoculum and ethanol productivity.

Keywords: Pharmaceutical Processes, Optimization, Dynamics and Control, Design Experiment, Batch Fermentation
Abstract: In a recent publication (Georgakis, 2013), it was shown that a data-driven model obtained through the proposed Design of Dynamic Experiments (DoDE) was able to accurately optimize a penicillin fermentation process without the use a knowledge-driven model. The resulting optimal operation, just after a set of experiments, is almost identical to the one obtained in (Riascos and Pinto, 2004) using the detailed model of the process by B&R (Bajpai and Reuss, 1980). Here we examine in silico whether a similar number of DoDE experiments will result in an equally accurate estimation of the optimal process operation of 32 other fermentation processes. Only very few of the 32 fermentations require additional experiments to obtain a satisfactory process optimization through a more accurate data-driven model. Furthermore, we examine two different time-evolving domains, A and B, within which the substrate inflow is varied. The obtained optimal operation in domain B is always better than that in domain A; sometimes by as much as 280%.

Keywords: Modelling and Identification, Systems Biology
Abstract: The phenomenon by which textit{Escherichia coli (E. coli)} move towards or away from a chemical by altering its swimming pattern is termed as Chemotaxis. The binding of ligands to trans-membrane receptors present on the cell surface results in a series of intra-cellular reactions that ultimately controls the bias of the flagella motors and in turn alters the swimming pattern. A similar phenotypic response has been observed in gradients of oxygen, termed as aerotaxis, though the intra-cellular pathway responsible for the effect is not understood. In this paper, we propose a simple model based on a generic adaptation mechanism that captures the cell's response to a time varying concentration of oxygen. Our model incorporates both the mechanisms of aerotaxis and chemotaxis and predicts the motion of a cell in a gradient of Methyl Aspartate. The predictions are in good agreement with the measurements of drift velocity in controlled gradients of Methyl Aspartate.

Keywords: Systems Biology, Dynamics and Control
Abstract: Heterogeneity is a common property of biological signalling systems. In most cases, heterogeneity is described qualitatively, only for some classes of responses have qualitative measures been proposed. For cell death signalling, this paper is the first to propose a quantification of heterogeneity. The challenge hereby is the dual aspect of heterogeneity. First, only part of the cell population may die while the others survive a specific death stimulus. Second, the time of death can vary from cell to cell.

The proposed heterogeneity measure is based on an L1 measure of the deviation between a homogeneous response and the population cumulative density function, on a nonlinearly scaled time. This measure allows for a quantitative study of the dependency of the heterogeneity of the responses to different stimulus doses or parametric variations. This will for example enable sensitivity analyses of heterogeneity.

The heterogeneity measure is illustrated by applying it to two different published cell ensemble models of apoptosis signalling, each having approximately 50 states and over 100 kinetic parameters. This analysis reveals that heterogeneity is more pronounced at an intermediate range of doses. In other words, high doses or low ones yield more homogeneous responses in the cell population.

Keywords: Systems Biology, Modelling and Identification, Environmental Processes (Wastewater, Bioremediation)
Abstract: We propose a dynamical model depicting nitrogen (N2) xation (diazotrophy) in a unicellular cyanobacterium, Crocosphaera watsonii, grown under light limitation and obligate diazotrophy. In this model, intracellular carbon and nitrogen are both divided into a functional pool and a storage pool. An internal pool that explicitly describes the nitrogenase enzyme is also added. The model is successfully validated with continuous culture experiments of C. watsonii under three light regimes, indicating that proposed mechanisms accurately reproduce the growth dynamics of this organism under various light environments. Then, a series of model simulations is run for a range of light regimes with dierent photoperiods and daily light doses. Results reveal how nitrogen and carbon are coupled, through the diel cycle, with nitrogenase dynamics, whose activity is constrained by the light regime. We nally identify optimal productivity conditions.

Keywords: Bioinformatics, Integrated Bioprocessing
Abstract: DNA separation is required to be rapid to be a useful component in DNA analysis devices. Different microfluidic device structures can be exploited to separate DNA with high throughput. We presents a framework for determining the optimal microfluidic device structure for rapid DNA separation through solving a nonlinear programming problem. Optimally designed spiral and serpentine microfluidic device configurations are shown to give comparable results for separating up to 425 bases of DNA using the micelle end-labeled free solution electrophoresis technique. The minimum run time for the serpentine microfluidic device configuration separating up to 425 bases of DNA is 5.1 minutes.

Keywords: Modelling and Identification, Systems Biology
Abstract: Interferon is a key signalling molecule in our innate immune system and is triggered in response to viral infections. Interferon signalling results in the expression of several hundred genes which together induce an antiviral state in cells and prevent the progression of disease. Yet, infections by viruses such as HIV and hepatitis C virus result in lasting, chronic infection. Understanding when interferon fails to clear viral infections may lead to the design of novel and potent strategies of intervention. We analyzed the interferon signalling network in the presence of viral infections, which can suppress the network at various steps, and discovered that the network exhibits bistability. In one steady state, viral levels are controlled and interferon levels are high. In the other, viral levels are high and interferon is suppressed. We found this bistability to be a general feature of the interferon network modulated by viral infections. Due to intrinsic variations, cells can admit either one or both the steady states. Cells that admit the second steady state alone are refractory to interferon. When the latter cells are higher in proportion than a critical value, interferon fails and the infection becomes chronic. In this talk, the above analysis of the interferon signalling network will be presented along with strategies to exploit the insights gained for targeted intervention.

Keywords: Dynamics and Control, Modelling and Identification, Environmental Processes (Wastewater, Bioremediation)
Abstract: This contribution explores the use of diagnosis and control modules based on fuzzy set theory and logic for bioreactor monitoring and control. With this aim, two independent modules were used jointly to carry out first the diagnosis of the state of the system and then use transfer this information to control the reactor. The separation in diagnosis and control allowed a more intuitive design of the membership functions and the production rules. Hence, the resulting diagnosis-control module is simple to tune, update and maintain while providing a good control performance. In particular the diagnosis-control system was designed for a complete autotrophic nitrogen removal process. The whole module is evaluated by dynamic simulation. Additionally, the diagnosis tool was demonstrated by analysis 100 days of experimental data.

Keywords: Dynamics and Control
Abstract: This paper presents a simple procedure to design fractional order controller based on synthesis method for the biochemical reactor. The biochemical reactor process exhibits high degree of non linearity. The process parameters will be be varying during the process of fermentation. Hence an attempt is made to design robust PI controller for the biochemical reactor to achieve high steady state productivity.

Keywords: Modelling and Identification, Systems Biology, Environmental Processes (Wastewater, Bioremediation)
Abstract: Natural gas contains a small percentage of hydrogen sulphide which should be recovered, before the fuel distribution, since it is a toxic and corrosive component. Sulphur recovery is usually done from a catalysis process where an adsorbent material removes the sulphide from the natural gas. Since this process works continuously with large input flow rates, the catalyst reaches a saturation state and it should be substituted and/or reactivated. Chemical are the most used method for the reactivation of such catalysts; nevertheless alternative methods have been explored from some years ago. In this work a biological method is studied in order to remove the sulphur from an exhaust catalyst used in the Pemex (the Company which manage oil and natural gas in Mexico) sulphur recovery plants. A kinetic analysis is done and a mathematical model, including parameter identification, for Thiosphaera pantotropha is presented. Experimental data from a laboratory set-up are employed to do this study. A removing of around 80% of the sulphide from the exhaust catalyst is obtained from preliminary laboratory experiments. These results are comparable with the reported in the literature and can be used for a future scaling procedure.

Keywords: Parameter and State Estimation, Systems Biology, Environmental Processes (Wastewater, Bioremediation)
Abstract: A based neural networks state observer to estimate biomass, substrate and methane in a continuous anaerobic reactor is introduced in this paper. The observer is designed from a recurrent high order neural network with a hyperbolic tangent as activation function and an extended Kalman filter as learning algorithm. The observer structure is validated via simulations and using experimental data obtained from an anaerobic continuous stirred tank at lab scale. This prototype is used to treat real slaughterhouse wastewater and it is operated in continuous mode. The obtained results show that the proposed observer is able to reproduce adequately the biomethane production and the substrate (related to chemical oxygen demand) in the methanogenesis stage; besides, methanogenic bacteria are also well estimated but some modifications are required in order to reach better results.

Keywords: Microbial Technology, Modelling and Identification, Metabolic Engineering
Abstract: Saccharomyces cerevisiae, a eukaryotic model organism, is considered the ideal host for microbial production of plant secondary metabolites such as polyketides and alkaloids. However, industrial scale production of these valuable products using S. cerevisiae is limited by the availability of their precursor, aromatic amino acid tyrosine. Here, we describe a framework which uses a combination of computational modeling techniques to design an in silico metabolic engineering strategy that improves the flux through the aromatic amino acid pathway (shikimate pathway) in S. cerevisiae. The predicted yeast strain can be used as a platform strain for production of any heterologous products which require tyrosine, or any other aromatic amino acid pathway metabolites as precursors. The initial genome-scale strain design was performed using steady-state constraint-based modeling methods, Optknock and GDLS. The resulting design required deletion of multiple genes and was difficult to validate experimentally. In order to obtain an experimentally feasible design, a small-scale kinetic model was developed using Ensemble Modeling, and was used to prioritize the knockouts predicted by steady-state models for experimental validation.

Keywords: Systems Biology, Bioinformatics, Dynamics and Control
Abstract: MicroRNAs (miRNAs) are small non-coding RNAs that regulate many genes at transcriptional and post-transcriptional level. Many miRNAs are involved in cancer pathways. A number of computational approaches are available for detecting static miRNA activity from its target gene expression data but none predicts temporal miRNA activity. In the current study we combined miRNA-gene interactions with gene expression data to predict temporal miRNA activity. We used network component analysis (NCA) to build a dynamic network of epidermal growth factor receptor (EGFR) signaling involving key miRNAs and its target genes. We clustered the genes in the context of predicted regulatory strength matrix and it facilitated the genes into clusters which explain coordinated and sequential regulation of miRNAs. Finally, the pathway analysis of differentially expressed targets revealed that EGFR is strongly involved in many cancer related pathways.