Sloppy control in bacterial growth and division

In balanced exponential growth, bacteria maintain statistical stability despite a high level of noise. Cell size, inter-division time, and other properties are stably distributed over hundreds of division cycles. This "homeostasis" has fascinated scientists for decades. Recent technologies have resurged interest and advanced this field tremendously, yet much remains to be understood. In particular, the control of multiple coupled variables simultaneously poses a problem in both data and model analysis: It is not clear, a-priori, which variables are actually controlled and which are stabilized through coupling. In this talk I will present an empirical approach to address this problem, based on the notion of robustness to perturbations around a setpoint commonly used in engineering theory. Applying it to single-cell bacterial data, we find a hierarchy of control, from tightly controlled variables whose setpoints are stiff, to sloppy variables whose setpoint drifts with the environment. This “sloppy control” hierarchy (a term borrowed from sloppy parameter dependence) is reflected geometrically as a control manifold in the space of variables. Setpoints span a wide range of values within the manifold, whereas out-of-manifold deviations are constrained. Several surprises appear, as we identify quantities that are traditionally considered regulated to be at the sloppy end of the spectrum. Our work offers a data-driven approach for identifying control variables in a multidimensional system that can be generalized to other systems.