A descriptive marker gene approach to single-cell pseudotime inference.

Area of Science:

• Computational Biology
• Genomics
• Systems Biology

Background:

• Single-cell gene expression data provides static snapshots of cellular states.
• Pseudotime estimation aims to reconstruct developmental trajectories from this data.
• Current methods often lack interpretability regarding gene behavior along trajectories.

Purpose of the Study:

• To introduce Ouija, a novel Bayesian approach for pseudotime inference.
• To enable interpretable modeling of gene behavior along inferred trajectories.
• To validate Ouija's performance against existing transcriptome-wide methods.

Main Methods:

• Utilizes a small set of marker genes for pseudotime inference.
• Models gene behavior as switch-like or transient along the trajectory.
• Employs an orthogonal Bayesian framework.

Main Results:

• Ouija recovers pseudotimes consistent with whole-transcriptome approaches.
• The method provides interpretable parameters for individual gene behavior.
• Identifies differences in gene regulation timing and discrete cell states ('metastable' states).

Conclusions:

• Ouija offers an interpretable alternative for pseudotime estimation in single-cell genomics.
• The approach facilitates understanding of gene dynamics during cellular transitions.
• Enables the discovery of biologically relevant cell types within continuous trajectories.