TIVelo: RNA velocity estimation leveraging cluster-level trajectory inference.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• RNA velocity inference is crucial for studying cell dynamics, including development, differentiation, and disease.
• Current methods often rely on ordinary differential equation (ODE) assumptions, limiting their ability to capture complex transcriptomic patterns.
• This limitation can lead to inaccuracies in gene expression velocity estimation.

Purpose of the Study:

• To introduce TIVelo, a novel RNA velocity estimation approach.
• To overcome the limitations of ODE-dependent methods in capturing complex transcriptomic patterns.
• To improve the accuracy and robustness of RNA velocity inference.

Main Methods:

• TIVelo determines velocity direction at the cell cluster level using trajectory inference.
• It calculates an orientation score to infer cluster-level direction without explicit ODE assumptions.
• Velocity is then estimated for individual cells based on the inferred cluster-level direction.

Main Results:

• TIVelo effectively captures complex transcriptional patterns by avoiding strict ODE assumptions.
• The method demonstrates improved velocity estimation for genes not adhering to simple ODE models.
• Validation across 16 real datasets showed TIVelo's effectiveness compared to six benchmarking methods.

Conclusions:

• TIVelo offers a more robust and accurate approach to RNA velocity inference.
• The method's ability to handle complex transcriptomic patterns enhances its utility in biological research.
• TIVelo provides a valuable tool for understanding cell development, differentiation, and disease progression.