Research topics
Our main interests are in:
- Synthetic genetic circuits for control of cell population variability and metabolic burden
- Multi-criteria optimization-based tuning of synthetic genetic circuits
- Possibilistic methods for Metabolic Flux Analysis and Flux Balance Analysis
- Metabolic models and regulation
- Control of specific growth rate in fed-batch bioreactors
- Sliding-mode reference conditioning to cope with metabolic overflow
- Observers for specific growth rate in fed-batch bioreactors
Research Lines
In this work we first explore the variability in LuxI and AHL species concentrations in a synthetic genetic circuit implementing a feedback loop and cell-to-cell communication.
Four different configurations: open and closed loop without and with cell to cell communication, allows us to see how feedback loop with cell to cell communication effectively reduces the variability across the population of cells.
To have a more analytical insight and a quantitative measure of how much the synthetic circuit improves the variability along the population, we approximate the sigmoidal repressible promoter response with a piecewise linear function with saturation, and we lump all the variability into the tightness of the promoters transcription rate.
With all this, we present a framework where under certain situation, the systems converge into the linear region of the controller, and then we obtain a closed expression for the coefficient of variation of the steady state distribution. This results in the possibility of fine-tune the parameters of the controller to obtain a desired output variability in the protein concentration.
Model based design plays a fundamental role in the new era of synthetic biology where complexity comes to stay. Modularity, i.e. using biological parts and interconnecting them to build new and more complex devices is the way to go. In this context, mathematical models have been used to generate predictions of the behavior of the designed device.
Now, designers not only want the ability of behavior prediction but also want aids and help on the design and selection of biological parts. However, using a model to determine crisp values for parameters of the involved parts is not a realistic approach, since uncertainties are ubiquitous to biology, and the characterization of biological parts is not exempt of it.
Here, we study the possibility of a multi objective optimization framework to get a model based set of guidelines for the selection of biological parts parameters to build a given device. The design criteria are encoded in the formulation of the objectives and optimization problem itself. As a result, the designer obtains qualitative regions/intervals of parameters giving rise to the predefined behavior.
A natural property of the framework is that is easy to analyze the impact of context on the synthetic devices to be designed, by just incorporating some information on the relationship of the device with the context. In general, this means we only need to know where we are connecting our device. Including this information in the optimization problem, we obtain a qualitative region of parameters taking into account the effect of the context on our device. Then, we can find a larger region of parameters if we allow for a set of desirable output performances, obtaining more robust guidelines for the design of synthetic devices.
Constraint-based models allow the calculation of the metabolic flux states that can be exhibited by cells, standing out as a powerful analytical tool, but they do not determine which of these
are likely tobe existing under given circumstances. Typical methods to perform these predictions are (a) flux balance analysis, which is based on the assumption that cell behaviour is optimal, and (b) metabolic flux analysis, which combines the model with experimental measurements.
In this research line we develop a possibilistic framework to perform metabolic flux estimations using a constraint-based model and a set of measurements. The methodology is able to handle inconsistencies, by considering sensors errors and model imprecision, to provide rich and reliable flux estimations. The methodology can be cast as linear programming problems, able to handle thousands of variables with efficiency, so it is suitable to deal with large-scale networks. Moreover, the possibilistic estimation does not attempt necessarily to predict the actual fluxes with precision, but rather to exploit the available data - even if those are scarce - to distinguish possible from impossible flux states in a gradual way.
Cellular metabolism consists of thousands of enzymatic reactions that extract energy from nutrients and assemble macromolecules that are necessary for cell survival. The size and complexity of metabolic networks can often obscure the roles of functionally related reactions on the overall distribution of metabolic fluxes. Moreover, different environmental contexts force the cell to reshape the distribution of fluxes to accommodate for changes in nutritional conditions subject to chemical and thermodynamical constraints.
Here we present a systematic procedure to obtain informative and realistic network descriptions of metabolism that incorporate different layers of information and are tailored to different environmental contexts. We build a family of networks that incorporate the directionality of fluxes, stoichiometry of reactions, and optimal flux distributions predicted by Flux Balance Analysis (FBA). These models can then be analysed using the rich repertoire of techniques from network science.
Fed-batch processes, extensively used in the biotechnological industry, present a large number of obstacles to control engineers. The control designer must deal with complex dynamic behavior of microorganisms, strong modeling approximations, external disturbances, nonlinear and even inherently unstable dynamics, scarce on-line measurements of most representative variables, etc.
From a biological standpoint, the ideal control of a biotechnological process would achieve microorganisms to reach a (possibly time-varying) metabolic state at which their physiological behavior is appropriate for the desired goals: e.g. production of a given metabolite. To that end, control of fermentation processes makes use of available measured or estimated variables that somehow can be related to the cell metabolic state as a function of nutrients supply. In this respect, cell growth underlies many key cellular and developmental processes. Thus, the desired microorganism metabolic states are usually strongly related to growth rate, as key representative of the underlying metabolic processes. Consecuently, its control is the underlying main problem in many cases.
In this research line, we develop specific growth rate control strategies based on the minimal model paradigm, requiring only biomass and volume measurement along with some bounds on the reaction rate. The controller has the structure of a partial state feed-back with adjustable gain. An integral-proportional control algorithm is designed to adjust this gain. First, a nonlinear integral action that makes invariant a goal manifold defined by a reference model dynamics is developed. Then, a proportional output error feed-back is incorporated to the control law with the aim of fastening convergence.
In many biotechnological processes, the optimal productivity corresponds to operating at critical substrate concentration. The problem, then, consists of maximizing the feeding rate compatible with the critical constraint, so as to avoid overflow metabolism. This value may be unknown and may change from experiment to experiment and from strain to strain, and even in the same experiment due to changing environmental and/or process conditions. In previous works different strategies to cope with this problem have been applied to microorganisms of industrial interest, such as E. coli and S. cerevisiae. Thus, probing strategies have been used in fedbatch bioreactors to operate close to their maximum oxygen transfer rate while avoiding acetate accumulation in the first case. In the fed-batch fermentation of S. cerevisiae a small amount of ethanol is allowed to be present in the culture, and the control problem in one of regulating the ethanol concentration a a given low reference value.
Here an approach based on sliding mode reference conditioning is proposed to drive the system to a maximum specific growth rate compatible with a given constraint (e.g. ethanol concentration lower than a given threshold). It is shown how this approach is robust with respect to uncertainties in the process dynamics and with respect to unknown perturbations affecting the critical point.
In this research line modified second-order sliding mode observers are designed for signal reconstruction in bioreactors. They have been specifically designed to estimate the specific growth rate of microorganisms based on biomass measurement. All batch, fed-batch processes, and continuous process applications are considered. The observers design design is not based on any model for the kinetics of the reaction, which may be monotonic or not. Just an upper-bound on its time derivative is required to tune the observer parameters.
These observers are equivalent, after some coordinate and time scale transformations, to the so-called super-twisting sliding algorithm, thereby inheriting its attractive features. In contrast with continuous observers, perfect tracking after finite convergence time can be achieved in the absence of noise, whereas chattering caused by noise is substantially reduced in comparison with con-ventional sliding observers. This theoretical property, i.e. finite time convergence, is very attractive in real-world control applications since the separation principle can be applied to design observer and controller independently.
Previous projects
MULTISYSBIO is a research project leaded by the Control of Complex Systems Group, involving people from GCSC-UPV, GIEM-UPV and IIM-CSIC, and aiming at a multi-scale modelling approach to systems biology: application to bioprocess monitoring, optimization and control. | |
BIOCONTROL is a research and collaboration project between the Universidad Politécnica de València (Spain) and the Universidad Nacional de La Plata (Argentina) aiming the modelling, sensorization and control of bioreactors. |