Classifying cell fate transitions in high-dimensional gene regulatory networks

In many animals, like mice and humans, cells to perform a wide range of roles. For example, alveolar cells in the lung are responsible for gas exchange, cells in the intestinal lining absorb nutrients, and neurons quickly transmit signals to one another. All of these specialized tissues are important for maintaining homeostasis. Characterizing how specialized tissues develop from stem cells, or how cells transition from one function to another, is an exciting question for both biologists and physicists interested in emergent behavior. There has been much progress towards building a theory of transitions using the mathematics of bifurcations and effective landscapes. However, connecting the concept of low-dimensional landscapes to the reality of high-dimensional gene regulatory networks remains an open challenge. In this talk, I will describe an approach to bridging this gap that is built on the framework of spin systems and, more specifically, Hopfield networks. Using generalized order parameters and existing high-dimensional scRNA-seq data, we show various classes of phase transitions in experimental data. We also discuss the role of annealing in determining the class of phase transition, demonstrating the effectiveness of describing complex biological systems using the tools of statistical physics.