Quantifying the Single-Cell Morphological Landscape of Cellular Transdifferentiation through Force Field...

Area of Science:

• Single-cell biology
• Computational biology
• Biophysics

Background:

• Transcriptomics advanced cell behavior understanding, but quantitative multi-omic models, especially for cell morphology, are lacking.
• Cell-specific velocity information is crucial for analyzing cellular dynamics and thermodynamics but is often unavailable.
• Analyzing cell fate transitions requires understanding the underlying forces driving morphological changes.

Purpose of the Study:

• To develop a novel machine learning approach for reconstructing cellular force fields from sparse single-cell imaging data.
• To quantitatively model cell morphology changes during transdifferentiation.
• To extend the landscape and flux framework to non-steady-state conditions for dynamic cellular processes.

Main Methods:

• Captured snapshots of fibroblast-to-neuron transdifferentiation.
• Developed a machine learning method to reconstruct the force field driving morphological changes.
• Decomposed the force field into flow flux and potential gradient components, adapting to non-steady-state conditions.

Main Results:

• The reconstructed force field accurately captured the cell morphological landscape during fate switching.
• The study revealed the influence of noise on cell state transitions.
• The developed method provides a quantitative framework for analyzing cell morphology dynamics.

Conclusions:

• The novel machine learning approach successfully reconstructs cellular force fields from sparse imaging data.
• This method enables the analysis of cell fate transitions and dynamics without direct velocity measurements.
• The framework is applicable to various single-cell multi-omic datasets lacking inherent velocity information.