A computational method for inferring reversible and irreversible single cell state transitions

A major question in biology is to understand patterns of cell state transitions and their governing regulatory mechanisms. In this study, we devise a new computational method, named state transition inference using cross-cell correlations (STICCC), for inferring reversible and irreversible cell state transitions by using single cell gene expression data and a gene regulatory network. The key assumption of the method is that there are time delays between regulators' activity and targets' expression, and such a feature can be exploited to infer directions of cell state transitions. We apply STICCC to both simulated and experimental single cell gene expression data and found that the inferred vector fields capture not only basins of attraction, but also irreversible fluxes. By associating gene regulatory relationship with systems' dynamical behaviors, STICCC reveals how specific genes and regulatory interactions contribute to the reversible and irreversible state transitions. STICCC will provide new insights into the gene regulatory mechanism of many biological processes involving cell state transitions.