scDIFF: automatic cell type annotation using scATAC-seq data by incorporating bulk-level genomic and epigenomic...

Area of Science:

• Genomics
• Computational Biology
• Epigenetics

Background:

• Single-cell assay for transposase-accessible chromatin sequencing (scATAC-seq) enables gene regulation analysis at single-cell resolution.
• Automatic cell type annotation for scATAC-seq data is crucial due to rapid data accumulation.
• Current methods often create artificial gene activity matrices, neglecting intrinsic scATAC-seq peak information.

Purpose of the Study:

• To develop an automated cell type annotation method for scATAC-seq data.
• To overcome limitations of existing methods that rely on artificial gene activity matrices.
• To leverage intrinsic information within scATAC-seq peaks for improved annotation.

Main Methods:

• Introduced scDIFF, a diffusive transformer-based computational method.
• Integrated bulk-level genomic and epigenomic information with scATAC-seq data.
• Avoided the creation of artificial gene activity matrices for annotation.

Main Results:

• scDIFF demonstrated superior and consistent performance across 46 benchmarking datasets.
• The method outperformed state-of-the-art cell type annotation techniques.
• Performance was robust across different sequencing platforms.

Conclusions:

• scDIFF provides an effective approach for cell type annotation using scATAC-seq data.
• The method's ability to directly utilize peak information enhances annotation accuracy.
• scDIFF offers a valuable, well-documented, and freely available tool for the research community.