New control theoretic frameworks for engineering biological control

Synthetic and engineering biology has the potential to provide solutions to many contemporary issues. Engineering biological controllers in living cells confers them with desirable functions such as disturbance rejection, "synthetic homeostasis", noise filtering and improved robustness. However, numerous theoretical challenges remain to the engineering of these controllers due to the inherent properties (such as stochasticity, lack of modularity and nonlinearities) of biological systems.

In this project, we will develop systems and control theoretic tools that enable these biological phenomena to be accounted for during biological controller design.

On the challenges related to nonlinearities, we will develop a new mapping approach to facilitate the transformation of nonlinear processes into linear models to enable classical linear control theory to be applied to biological controller design. We will then use this mapping approach to identify parameters and controller designs that reproduce their dynamics for nonlinear functions.

We will extend the approach by supplementing it with mechanistic models of gene expression and explore how circuit topologies can be modified to linearise the responses.

On the challenges related to lack of modularity, we will develop new design frameworks that addresses two sources of modularity failure: retroactivity and resource competition. Combining genetic circuits allows greater complexity and more information processing capacity but also leads to the dynamics of upstream process becoming impacted by downstream processes (retroactivity), whilst the finite synthesis capacity of cells results in nonlinear indirect interactions between genetic modules (resource competition).

We will then explore the potential of feedback control to ameliorate the impact of both effects both individually and concurrently.

On the challenges related to stochasticity, we will adapt and apply tools from robust control to assess controller designs to parametric uncertainty and develop stochastic simulations to test the impact of molecular noise on the designed controllers.

We will work with experimental collaborators to design real-world network motifs for the robust control of protein production.