Global parameter estimation methods for stochastic biochemical systems.

Area of Science:

• Systems Biology
• Computational Biology
• Biophysics

Background:

• Stochastic models, such as the chemical master equation, are vital for studying cellular processes with low molecule counts.
• Accurate kinetic parameters are essential for analyzing system properties and predicting genetic perturbation effects.
• Parameter estimation for stochastic models is challenging but increasingly important due to advances in single-molecule measurement technologies.

Purpose of the Study:

• Develop practical and effective methods for estimating kinetic model parameters.
• Address parameter estimation for the chemical master equation and other stochastic models.
• Utilize both single-cell and cell population experimental data.

Main Methods:

• Propose three parameter estimation methods: maximum likelihood and two density function distance methods (probability and cumulative density functions).
• Account for finite sampling effects in Monte Carlo simulations when analyzing state density functions.
• Apply methods to three practical case studies.

Main Results:

• Maximum likelihood method effectively handles low replicate measurements.
• Density function distance methods, especially cumulative density function distance, offer robust parameter estimation with higher accuracy.
• Cumulative density function distance estimation performs well even in systems exhibiting multimodality.

Conclusions:

• The developed methodologies provide an effective and practical approach for estimating kinetic parameters of stochastic systems.
• The methods are applicable to both sparse and dense cell population data.
• Parameter identifiability remains a challenge, with not all parameters being accurately estimable from available data.