Abstract

The derivation of hard performance bounds in the presence of noisy feedback has been extensively addressed by control theorists over the last century, with many results going back to the early work of Wiener, Bode and Kalman. In a biochemical reaction network, however, where the noise is associated with the spontaneous births and deaths of individual molecules, many of the more conventional methodologies come to a halt, and need to be appropriately refined and extended. The approach in this talk is to address such complications by quantifying the way these spontaneous fluctuations restrict the ability to transmit information reliably, i.e. they lead to communication channels with finite capacity in the sense of Shannon, which can be explicitly calculated. We show that such restrictions in information transmission fundamentally limit the ability to suppress fluctuations in molecular concentrations, thus leading to hard bounds for noise suppression that hold for arbitrary feedback policies. The limits derived challenge conventional beliefs about biochemical accuracy, and show how an interface between control and information theory can be used to rigorously analyze poorly characterized biological systems.

Parts of the results that will be presented are in press for publication in Nature as a full research article.