Modeling Variability in Populations of Cells Using Approximated Multivariate Distributions.

Area of Science:

• Computational Biology
• Systems Biology
• Biophysics

Background:

• Cell populations exhibit inherent variability in species concentrations, crucial for understanding population dynamics.
• Accurately modeling this variability is challenging due to the curse of dimensionality in high-dimensional probability distributions.

Purpose of the Study:

• To develop novel methods for approximating and tracking the joint probability distribution of species in cell populations.
• To improve the accuracy and efficiency of modeling biological pathways with inherent variability.

Main Methods:

• Utilized Chow-Liu tree representations to approximate joint probability distributions, capturing correlations between species.
• Developed and evaluated a new approximate inference algorithm for tracking the evolution of these distributions over time.
• Compared methods against classical approximations and existing inference algorithms using data from Ordinary Differential Equations (ODEs) models.

Main Results:

• The proposed Chow-Liu tree approximation scheme demonstrated higher accuracy than existing methods for modeling probability distributions from biopathways.
• The new inference algorithm, combined with Chow-Liu trees, achieved high accuracy with minimal computational overhead.
• Experimental results validated the effectiveness of the developed methods for complex biological pathway dynamics.

Conclusions:

• Chow-Liu tree approximations offer a more accurate approach to modeling species variability in cell populations compared to traditional methods.
• The novel inference algorithm provides an efficient and accurate solution for tracking population dynamics in biological systems.
• This work advances the computational modeling of complex biological systems by addressing the challenge of high-dimensional probability distributions.