Cell2fate infers RNA velocity modules to improve cell fate prediction

Area of Science:

• Computational Biology
• Genomics
• Systems Biology

Background:

• RNA velocity infers cell state dynamics from spliced and unspliced RNA counts.
• Current models face accuracy limitations due to biophysical simplifications and numerical approximations.

Purpose of the Study:

• To introduce cell2fate, a novel RNA velocity framework based on linearized ODEs and Bayesian inference.
• To improve the accuracy and interpretability of transcriptional dynamics inference.

Main Methods:

• Linearization of RNA velocity ordinary differential equations (ODEs).
• Fully Bayesian inference for solving the ODEs.
• Decomposition of RNA velocity solutions into interpretable modules.

Main Results:

• cell2fate demonstrates enhanced accuracy in reconstructing complex transcriptional dynamics.
• The method effectively identifies weak dynamical signals in rare and mature cell types.
• cell2fate provides a biophysical link between RNA velocity and dimensionality reduction techniques.

Conclusions:

• cell2fate offers a more accurate and interpretable approach to RNA velocity analysis.
• The framework is powerful for studying complex cellular trajectories and developmental processes.
• Application to human brain development reveals spatial mapping of transcriptional dynamics.