Comprehensive benchmarking of RNA velocity methods across single-cell datasets

Area of Science:

• Computational Biology
• Single-Cell Genomics
• Systems Biology

Background:

• RNA velocity analysis is key for understanding cellular dynamics from single-cell RNA sequencing data.
• The growing number of computational tools necessitates systematic evaluation.
• Existing comparisons lack scope and standardized metrics, hindering method selection.

Purpose of the Study:

• To establish a comprehensive benchmark for RNA velocity computational methods.
• To evaluate method performance across diverse biological and technical scenarios.
• To provide task-aware guidance for selecting appropriate RNA velocity tools.

Main Methods:

• Benchmarked 30 distinct RNA velocity methods from 19 tools.
• Evaluated 25 RNA-only methods on directional consistency, temporal precision, negative control robustness, and sequencing depth stability.
• Assessed 5 multimodal-enhanced methods on multimodal integration using 34 datasets (26 real-world, 8 simulated).

Main Results:

• Identified a trade-off between directional consistency and negative control robustness.
• Observed distinct performance patterns based on temporal modeling strategies.
• Found variability influenced by sequencing depth and quantification choices.
• Highlighted gaps in gene dependence modeling, temporal inference, and multimodal architectures.

Conclusions:

• Established a unified framework for RNA velocity method evaluation.
• Provided context-specific guidance for method selection, moving beyond single rankings.
• Emphasized the importance of aligning method choice with biological context and technical constraints.