On Noise and Filtering in Adaptive Signaling Networks

Abstract

Two different noise filtering strategies are identified and studied in a class of adaptive sensory systems. The high frequency noise is filtered by the Berg-Purcell time averaging scheme with the filtering carried out by the output decay process independent of the slow adaptation dynamics. The low frequency noise is reduced by adaptation and decreases as the feedback time shortens. Both filtering mechanisms introduce noises that are unfiltered and thus contribute to a significant internal noise floor. We show that both noise filtering mechanisms are utilized in the E. Coli chemotaxis pathway: the ligand binding noise is averaged by the response time and remains small due to its fast time scale, while the dominant signal noise, caused by the random cell motion in a gradient, is controlled by adaptation. We conclude that the chemotaxis pathway optimizes gradient sensing, strong response, and noise control in different time scales.