A likelihood-based Bayesian inference framework for the calibration of and selection between stochastic velocity-jump...

Area of Science:

• Biophysics
• Computational Biology
• Mathematical Modeling

Background:

• High-resolution spatio-temporal data enable tracking of individual motile entities.
• Calibrating mathematical models to experimental tracking data presents challenges due to discrete time steps and measurement noise.

Purpose of the Study:

• To develop a Bayesian inference framework for calibrating velocity-jump models to discrete-time, noisy single-agent motion data.
• To assess the framework's effectiveness in parameter recovery and model selection.

Main Methods:

• Developed a Bayesian inference framework leveraging approximate solutions to stochastic processes.
• Applied the framework to simulated data from two-state and three-state velocity-jump models.
• Utilized experimental data tracking mRNA transport in Drosophila neurons for model selection.

Main Results:

• Successfully recovered model parameters from simulated data for two-state and three-state models.
• Demonstrated effective calibration and selection between velocity-jump models using both simulated and experimental data.
• Showcased the framework's applicability to diverse motion processes.

Conclusions:

• The developed Bayesian framework is effective and efficient for calibrating and selecting velocity-jump models.
• The framework can be applied to analyze various single-agent motion processes, including biological transport.
• Advances in computational methods enhance the analysis of complex experimental tracking data.