Accurate RNA velocity estimation based on multibatch network reveals complex lineage in batch scRNA-seq data.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• RNA velocity analysis is crucial for understanding cell development using single-cell RNA sequencing (scRNA-seq).
• Existing RNA velocity methods struggle with batch effects, leading to inaccurate cell trajectory inference.
• Batch effects in spliced and unspliced RNA matrices compromise the reliability of velocity streams.

Purpose of the Study:

• To develop an innovative RNA velocity method, VeloVGI, that effectively corrects for batch effects.
• To improve the accuracy of cell development trajectory inference in scRNA-seq data.
• To enhance feature extraction for more robust velocity estimation.

Main Methods:

• VeloVGI utilizes an optimal transport (OT) and mutual nearest neighbor (MNN) approach to construct neighbors across batches.
• It incorporates graph structure into the encoder for improved feature extraction.
• The method refines velocity estimation building upon the VeloVI framework.

Main Results:

• VeloVGI successfully corrects for batch effects, overcoming limitations of existing methods.
• The incorporation of graph structure enhances feature extraction for more accurate velocity estimation.
• VeloVGI demonstrated superior performance compared to other methods across multiple datasets and biological scenarios.

Conclusions:

• VeloVGI offers a robust solution for batch effect correction in RNA velocity analysis.
• The method provides more accurate cell development trajectories by improving velocity stream estimation.
• VeloVGI represents a significant advancement for scRNA-seq data analysis, particularly in developmental studies.