Learning developmental mode dynamics from single-cell trajectories

Area of Science:

• Developmental Biology
• Biophysics
• Computational Biology

Background:

• Embryogenesis involves complex multicellular development driven by cell migration.
• High-resolution microscopy reveals intricate cell dynamics during development.
• Translating complex imaging data into predictive models remains a challenge.

Purpose of the Study:

• To develop a computational framework for learning continuum models from live-cell imaging data.
• To characterize developmental symmetry breaking in early zebrafish gastrulation.
• To infer interpretable hydrodynamic models of collective cell migration.

Main Methods:

• Utilized mode decomposition from physics and sparse dynamical systems inference.
• Applied harmonic basis functions to coarse-grain and compress cell trajectory data on curved surfaces.
• Mode-based model learning framework applied to zebrafish embryogenesis.

Main Results:

• Successfully learned a low-dimensional representation of collective cell dynamics.
• Enabled compact characterization of developmental symmetry breaking.
• Inferred a hydrodynamic model showing similarities to active Brownian particle dynamics.

Conclusions:

• Mode-based model learning provides a quantitative biophysical understanding of developmental processes.
• The framework is applicable to various developmental structure formation processes.
• Advances the study of cell migration in complex geometries during embryogenesis.