Generalized end-product feedback circuit senses high dimensional environmental fluctuations

Understanding computational capabilities of simple biological circuits, such as the regulatory circuits of single-cell organisms, remains an active area of research. Recent theoretical work has shown that a simple regulatory architecture based on end-product inhibition can exhibit predictive behavior by learning fluctuation statistics of one or two environmental parameters. Here we extend this analysis to higher dimensions. We show that as the number of inputs increases, a generalized version of the circuit can learn not only the dominant direction of fluctuations, as shown previously, but also the subdominant fluctuation modes.