Gene expression model inference from snapshot RNA data using Bayesian non-parametrics.

Area of Science:

• Computational Biology
• Systems Biology
• Molecular Biology

Background:

• Gene expression models are crucial for understanding cellular regulatory responses and single-cell transcriptional dynamics.
• Current computational methods require pre-specification of gene states and connectivity, limiting simultaneous inference of models and parameters.

Purpose of the Study:

• To develop a novel computational method for simultaneous Bayesian inference of gene expression models, including gene states, connectivities, and rate parameters, directly from single-molecule RNA counts.
• To address limitations of existing frameworks that necessitate pre-defined model structures.

Main Methods:

• Proposed a Bayesian non-parametric approach to learn full distributions over gene states, connectivities, and rate parameters.
• Treated gene expression models as random variables to propagate noise from RNA counts.
• Developed a self-consistent inference framework.

Main Results:

• Successfully demonstrated the method on the Escherichia coli lacZ and Saccharomyces cerevisiae STL1 pathways.
• Verified the robustness of the developed method using synthetic data.
• Enabled simultaneous and self-consistent inference of complex gene expression models.

Conclusions:

• The proposed method offers a significant advancement in inferring gene expression models from single-cell RNA data.
• This approach provides a more comprehensive understanding of cellular regulatory mechanisms by learning model components simultaneously.
• The method's robustness across different biological systems and synthetic data suggests broad applicability.