A qualitative analysis of ubiquitous regulatory motifs in Saccharomyces cerevisiae genetic networks

This work examines bistability and multistability within a Recurrent Neural Network model (RNN) for a 2-node and 3-node system under many different regulation scenarios. We determine parameter regions where there is bistability, multistability, or other stable modes in the expression states of the systems described by this network model. Our results show that although bistability can be generated with autoregulation it is also the case that both autorepression or no autoregulation can yield bistability as long as a sigmoidal behavior is present. Additionally, our results show the importance of considering more than a single connection when inferring a network as the observed biological result is averaged over many outcomes, which has implications for many algorithms that infer gene regulatory networks using the RNN models.