I will give a beginner’s-level introduction to Probabilistic Graphical Models in the context of a recent paper that refutes the PU1/GATA1 toggle switch paradigm of cell fate decisions in developmental biology. The authors argue that cell fate decisions are often timed based upon observations of live cell markers, but this could be flawed due to delays in development of such markers as well as microscope detection limits. Since real decision times are hidden, the authors suggest that correlations induced on lineages of single cells could instead be used to infer the actual cell fate decision times. They develop a sophisticated inference approach combining Probabilistic Graphical Models and Linear Programming to enumerate thousands of hidden lineage trees and infer the decision time point of early myeloid lineage choice. I will provide an overview of their method with the hope that it can lead to some interesting discussions and new ideas.
Relevant paper: https://www.nature.com/articles/s41467-018-05037-3