Benchmarking RNA velocity methods across 17 independent studies

Area of Science:

• Computational Biology
• Genomics
• Single-cell Analysis

Background:

• RNA velocity analysis is crucial for understanding cell differentiation dynamics.
• Existing computational methods for RNA velocity inference lack standardized best-practice guidelines.

Purpose of the Study:

• To benchmark existing RNA velocity inference methods.
• To provide evidence-based recommendations for method selection.

Main Methods:

• Evaluated 15 RNA velocity methods on 17 independent datasets.
• Assessed method performance based on accuracy, stability, and usability.
• Employed multiple validation strategies for robust evaluation.

Main Results:

• No single RNA velocity method demonstrated superior performance across all evaluated metrics.
• Variability in performance was observed, with some methods underperforming unexpectedly.
• Performance varied significantly depending on dataset characteristics and analytical goals.

Conclusions:

• The development of scenario-based best-practice suggestions is essential.
• Users should carefully select RNA velocity methods tailored to specific data types and research questions.
• Further research may be needed to refine existing methods and develop new ones.