Robust System Identification and Control of Cellular Processes
Controlling cellular processes presents unique challenges not often encountered in the control of traditional electrical, mechanical, or chemical systems. For example, mathematical models of cellular processes are typically nonlinear and uncertain. Furthermore, real-time, continuous feedback is generally not available and realizable control actions are limited by experimental techniques. As a result, there have been minimal efforts to apply control theory at the cellular level. An integrated engineering approach towards effective experiment design for predictably manipulating cellular responses will be discussed that employs robust controller design. An adaptive sparse grid based interpolation approach will be presented for partitioning an uncertain parameter space into unacceptable and acceptable subspaces. Robust parameters are identified as the most 'central interior' point of the acceptable subspaces so small perturbations about these values are less likely to lead to an unacceptable behavior. This method will initially be illustrated for robust model parameter identification with a standard mitogen-activated protein kinase (MAPK) cascade model prior to exploring its applicability to support robust controller design. Since nonlinear model predictive control (NMPC) has shown promise for controlling biomass production and batch cellular growth in bioreactor systems, an NMPC algorithm was devised that utilizes the robust sparse grid-based controller parameter selection to maximize the likelihood that an experimental strategy derived by the controller design will be successful in the laboratory environment. This robust NMPC approach was used to quantitatively design an experimental strategy (for an open-loop realization) that predictably manipulates the activation time course of the T-cell receptor activated MAPK, Erk. Preliminary experiments using a Jurkat T cell culture system evaluate the derived control strategy and provide insight to improve the supporting model and robust controller design approach. Ultimately, this engineering-control approach to quantitative experiment design is anticipated to advance our abilities to regulate cellular processes as well as improve our current understanding of them.