Learning developmental mode dynamics from single-cell trajectories

Over the last years, progress in high-resolution in-vivo imaging has provided unprecedented insight into the collective cell dynamics at different stages of embryogenesis. These rapid experimental advances pose the theoretical challenge of translating high-dimensional imaging data into predictive low-dimensional dynamical models that capture the essential principles governing developmental cell migration. Here, we have combined mode decomposition ideas that have proved successful in condensed matter and fluid physics with sparse dynamical systems inference to learn interpretable biophysical models from single-cell imaging data. Using zebrafish embryos as an example, we discuss how cell trajectory data can be coarse-grained and compressed. The resulting low-dimensional representation allows for quantitative comparisons of cellular and active Brownian particle dynamics, revealing similarities between their short term behaviors and emerging differences at longer time-scales.