Inference and multiscale model of epithelial-to-mesenchymal transition via single-cell transcriptomic data

Area of Science:

• Computational Biology
• Systems Biology
• Genomics

Background:

• Single-cell transcriptomics rapidly generates vast datasets, offering new avenues to study dynamic cellular processes.
• Understanding cell fate decisions, such as epithelial-to-mesenchymal transition (EMT), is critical in development and disease.

Purpose of the Study:

• To develop an integrative computational tool for analyzing cell fate transitions using single-cell transcriptomic data.
• To identify intermediate cell states (ICS) and the genes that regulate transitions during cell fate decisions.
• To explore the role of ICS in processes like EMT and their impact on population-level dynamics.

Main Methods:

• Combined unsupervised learning of single-cell transcriptomic data with multiscale mathematical modeling.
• Developed a novel single-cell gene regulatory network model.
• Applied the approach to twelve published single-cell EMT datasets from cancer and embryogenesis.

Main Results:

• Identified individual cells transitioning between states and inferred genes driving these changes.
• Revealed intermediate cell states (ICS) and their regulated transition trajectories.
• Uncovered the roles of ICS in adaptation, noise attenuation, and transition efficiency during EMT, including trade-off relationships.

Conclusions:

• The developed unsupervised learning method is broadly applicable to diverse single-cell transcriptomic datasets.
• The integrative, single-cell resolution approach can be adapted for studying other cell fate transition systems beyond EMT.