Simulation-based inference of differentiation trajectories from RNA velocity fields

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) enables the study of cellular heterogeneity and differentiation.
• RNA velocity offers insights into cell fate dynamics by analyzing transcriptional activity.
• Accurate trajectory inference is crucial for understanding developmental processes.

Purpose of the Study:

• To introduce Cytopath, a novel method for trajectory inference using RNA velocity.
• To evaluate Cytopath's performance in reconstructing complex differentiation trajectories.
• To compare Cytopath with existing trajectory inference methods.

Main Methods:

• Cytopath defines a Markov chain model based on single-cell RNA velocity data.
• It simulates an ensemble of possible differentiation trajectories.
• A consensus trajectory is constructed from the simulated ensemble.

Main Results:

• Cytopath successfully recapitulates topological and molecular characteristics of differentiation.
• The method reconstructs trajectories with bifurcated, circular, convergent, and mixed topologies.
• Cytopath's RNA velocity-based pseudotime correlates with actual elapsed time.
• Drawbacks in current trajectory inference methods were identified.

Conclusions:

• Cytopath provides a robust framework for RNA velocity-based trajectory inference.
• The method accurately captures complex differentiation dynamics from scRNA-seq data.
• Cytopath offers an advancement in understanding cell fate decisions.