Calibrating tissue level PDE models of ligand dynamics using single cell and spatial transcriptomics data.

Area of Science:

• Computational Biology
• Systems Biology
• Molecular Biology

Background:

• Cellular communication via ligand-mediated signaling is crucial for physiological processes.
• Reaction-diffusion models describe ligand dynamics but lack robust parameter calibration due to limited microenvironment data.
• Integrating diverse biological data into mechanistic models remains a challenge.

Purpose of the Study:

• To develop a computational pipeline for calibrating reaction-diffusion models using transcriptomics data.
• To leverage single-cell RNA sequencing and spatial transcriptomics for tissue-scale model parameter inference.
• To provide a unified framework for integrating modern transcriptomics data into mechanistic models.

Main Methods:

• Developed a computational pipeline integrating finite volume solvers, bioinformatics preprocessing, Approximate Bayesian Computation (ABC), and gradient-based optimization.
• Utilized open-source human skin datasets for case studies.
• Calibrated parameters for Transforming Growth Factor Beta isoforms.

Main Results:

• Successfully calibrated parameters governing Transforming Growth Factor Beta isoforms in human skin.
• Compared inferred spatial concentration fields with local cell type distributions.
• Benchmarking on synthetic datasets demonstrated the accuracy and benefits of combining ABC with gradient-based optimization.

Conclusions:

• Single-cell RNA sequencing and spatial transcriptomics offer a rich data source for calibrating tissue-scale mechanistic models.
• The developed pipeline provides a flexible and rigorous approach for parameter inference.
• This method enhances our ability to model cellular communication and tissue dynamics.