Geometric quantification of cell phenotype transition manifolds with information geometry

Area of Science:

• Computational Biology
• Systems Biology
• Genomics

Background:

• Cell phenotype transition (CPT) is vital for biological processes.
• Single-cell sequencing shows CPT dynamics occur on low-dimensional manifolds.
• Existing methods struggle to quantify these complex CPT manifolds.

Purpose of the Study:

• To develop a novel method for quantifying cell phenotype transition dynamics.
• To analyze the geometric properties of cell expression data.
• To identify critical transition points in cellular processes.

Main Methods:

• SCIM (single-cell information manifolds) approach utilizing information geometry.
• Embedding single-cell gene expression profiles as Gaussian distributions.
• Calculating coarse Ricci curvature and information velocity for each cell.

Main Results:

• Low-curvature regions in the data correspond to critical cell transitions.
• High information velocity aligns with low curvature, indicating geometric guidance.
• SCIM uncovers invariant features of cell phenotype transition manifolds.

Conclusions:

• SCIM provides a robust framework for quantifying cellular dynamics on transition manifolds.
• Geometric properties offer new insights into cell fate decisions.
• This method advances the understanding of developmental and biological processes.