Towards reliable quantification of cell state velocities.

Area of Science:

• Computational Biology
• Molecular Biology
• Genomics

Background:

• Single-cell RNA sequencing (scRNA-seq) provides snapshots of cellular states.
• Quantifying unspliced and spliced messenger RNA (mRNA) in scRNA-seq data offers temporal insights into transcriptional dynamics.
• Existing RNA velocity inference methods face theoretical and computational limitations.

Purpose of the Study:

• To identify and discuss theoretical and computational problems in current RNA velocity estimation methods.
• To propose novel solutions for enhancing data processing consistency, velocity inference, and visualization.
• To develop new computational tools for more reliable RNA velocity analysis.

Main Methods:

• Analysis of existing RNA velocity inference methodologies.
• Development of a detailed computational method (κ-velo) based on specific data assumptions.
• Development of a heuristic computational method (eco-velo) with different data assumptions.

Main Results:

• Identification of key challenges in current RNA velocity estimation.
• Proposal of solutions to improve data processing, inference, and visualization.
• Introduction of two new velocity analysis tools, κ-velo and eco-velo.

Conclusions:

• The proposed methods and tools aim to overcome limitations in current RNA velocity estimation.
• Accurate RNA velocity estimation can provide deeper insights into biological transcriptional dynamics.
• κ-velo and eco-velo offer distinct approaches for analyzing RNA velocity in scRNA-seq data.