Learning time-varying information flow from single-cell epithelial to mesenchymal transition data

Area of Science:

• Computational Biology
• Systems Biology
• Cellular Signaling

Background:

• Cellular regulatory networks are dynamic, reconfiguring in response to stimuli.
• Existing computational methods often model these networks as static, using single time-point data.
• This static approach limits understanding of dynamic cellular processes.

Purpose of the Study:

• To develop methods for learning dynamic protein-protein relationships from static single-cell data.
• To quantify and visualize time-varying interactions during cellular transitions.
• To predict the impact of drug perturbations on dynamic cellular processes like EMT.

Main Methods:

• Utilized mass cytometry data from a murine breast cancer cell line undergoing TGFß-induced epithelial-to-mesenchymal transition (EMT).
• Leveraged asynchronous EMT progression to construct a pseudotime EMT trajectory.
• Developed methods for visualizing and quantifying time-varying network edges along the trajectory.

Main Results:

• Successfully derived a pseudotime trajectory representing EMT progression.
• Introduced novel visualizations and quantification metrics for dynamic edge behavior.
• Developed a metric of edge dynamism capable of predicting drug perturbation effects on EMT.

Conclusions:

• Static single-cell data can be used to infer dynamic cellular regulatory network behavior.
• The developed methods provide new tools for analyzing dynamic biological processes.
• This approach enhances our ability to understand and predict responses to stimuli and drug treatments.