Topologically-based parameter inference for agent-based model selection from spatiotemporal cellular data.

TOPAZ accurately infers parameters and distinguishes between models in spatial single-cell data using topological analysis and Bayesian methods, with 100% success in simulations.

• Why it matters: Understanding cell population dynamics from complex spatiotemporal data is challenging, limiting insights into underlying biological mechanisms and intercellular interactions.
• What they did: The approach integrates topological data analysis with approximate Bayesian computation and model selection, utilizing persistent homology to capture spatial features from cellular trajectories.
• The result: This framework reliably recovers model parameters and differentiates models, enabling mechanistic inference and advancing analysis of spatial cellular behaviors with open-source tools.