TFvelo: gene regulation inspired RNA velocity estimation

Area of Science:

• Computational Biology
• Genomics
• Systems Biology

Background:

• RNA velocity, derived from single-cell RNA sequencing (scRNA-seq), predicts cell fate by analyzing the phase delay between unspliced and spliced mRNA.
• Existing RNA velocity models often struggle with insufficient dynamic signals from mRNA abundance, leading to poor model fits.
• Transcriptional regulation is a key driver of cellular dynamics, yet is underutilized in current RNA velocity frameworks.

Purpose of the Study:

• To develop a novel RNA velocity framework, TFvelo, that incorporates gene regulatory information.
• To expand RNA velocity modeling beyond splicing dynamics to various scRNA-seq datasets.
• To improve the accuracy and robustness of cell state prediction and trajectory inference.

Main Methods:

• TFvelo integrates gene regulatory network information with scRNA-seq data.
• The model analyzes RNA abundance, not solely relying on splicing information.
• Experiments were conducted on synthetic datasets and multiple real-world scRNA-seq datasets.

Main Results:

• TFvelo accurately fits gene dynamics in phase portraits.
• The method effectively infers cell pseudo-time and developmental trajectories.
• TFvelo demonstrates robust performance across diverse single-cell datasets.

Conclusions:

• TFvelo offers a novel and effective approach to RNA velocity analysis.
• The framework enhances the prediction of cell states and trajectories by leveraging transcriptional regulation.
• TFvelo provides a robust and accurate avenue for modeling RNA velocity in single-cell studies.