A model of transcriptional regulatory networks based on biases in the observed regulation rules

Control rules governing transciption of eukaryotic genes can be modeled as Boolean function, and these rules are strongly biased toward large numbers of “canalizing” inputs. The ensemble of networks with the observed canalizing bias predicts cells are in an ordered regime with convergent flow in transcription state space, a percolating subnetwork of genes fixed on or off an isolated islands of twinkling genes turning on or off, and a near power-law distribution of cascades of gene activity changes following perturbations. The data suggest that a given cell state or type can be represented as an attractor of transcriptional activity or flow over time.