scRDEN: single-cell dynamic gene rank differential expression network and robust trajectory inference

Area of Science:

• Computational Biology
• Genomics
• Developmental Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) enables precise gene expression analysis at the cellular level.
• Understanding cell fate determination requires analyzing dynamic regulatory alterations and inferring differentiation trajectories.
• Existing methods face challenges in comprehensively analyzing transcriptional heterogeneity and developmental paths.

Purpose of the Study:

• To introduce scRDEN, a robust framework for inferring cell subpopulations and differential expression networks.
• To analyze gene expression dynamics and regulatory alterations along differentiation trajectories.
• To provide insights into transcriptional regulation during cell differentiation and development.

Main Methods:

• scRDEN converts unstable gene expression values into stable gene-gene interactions (global features).
• It extracts differential expression order (network features) and integrates multi-dimensional reduction expression features.
• The framework analyzes gene expression, identifies cell subpopulations, and constructs differential expression networks.

Main Results:

• scRDEN successfully identified stable cell subpopulations with potential marker genes across five scRNA-seq datasets.
• It accurately measured transcriptional differences to establish rank differential expression networks along differentiation branches.
• Analysis revealed non-monotonic trends in network diversity and cluster coefficients, correlating with differentiation into stable functions.

Conclusions:

• scRDEN provides a robust framework for analyzing cell differentiation trajectories and regulatory mechanisms.
• The method offers novel insights into large-scale, multi-batch trajectory inference and transcriptional regulation.
• scRDEN demonstrated exceptional performance on complex datasets, including mouse dentate gyrus data.