Cell-mechanical parameter estimation from 1D cell trajectories using simulation-based inference.

Area of Science:

• Cell biology
• Computational modeling
• Biophysics

Background:

• Cell migration analysis relies on mathematical models, but parameter estimation is challenging.
• Stochastic and nonlinear models require advanced computational techniques for accurate parameterization.

Purpose of the Study:

• To apply simulation-based inference (SBI) for estimating cell-specific model parameters from cell trajectories.
• To utilize Bayesian inference and deep neural networks for analyzing cell migration data.

Main Methods:

• Automated time-lapse imaging and image recognition to record 1D single-cell trajectories.
• Training a deep neural density estimator using simulated trajectories from a mechanical cell migration model.
• Inferring probability distributions of model parameters using the trained neural network.

Main Results:

• Demonstrated SBI's efficacy in distinguishing between MCF-10A (non-cancerous) and MDA-MB-231 (cancerous) breast epithelial cells.
• Successfully identified the impact of Latrunculin A and Y-27632 inhibitors on cell migration model parameters.
• Validated the capability of SBI to uncover inhibitor effects without prior mechanistic assumptions.

Conclusions:

• SBI provides a powerful framework for analyzing cell migration data and estimating cell-specific parameters.
• The approach facilitates the creation of cell motility libraries and aids in evaluating refined migration models.
• This method offers new avenues for drug efficacy assessment and understanding cell migration mechanisms.