scMGCL: accurate and efficient integration representation of single-cell multi-omics data.

Area of Science:

• Computational Biology
• Genomics
• Systems Biology

Background:

• Single-cell multi-omics data integration is crucial for understanding cellular heterogeneity and disease.
• Integrating diverse data modalities like ATAC-seq and RNA-seq presents significant challenges.
• Existing methods struggle with robustly combining information from different single-cell assays.

Purpose of the Study:

• To present scMGCL, a novel graph contrastive learning framework for single-cell multi-omics data integration.
• To enable robust integration of single-cell ATAC-seq and RNA-seq data.
• To learn shared representations while preserving modality-specific features.

Main Methods:

• scMGCL utilizes a graph contrastive learning framework.
• It employs self-supervised learning on cell-cell similarity graphs.
• Cross-modality graph structures are used as augmentations for each other.

Main Results:

• scMGCL outperforms state-of-the-art methods in cell-type clustering and label transfer.
• It demonstrates superior preservation of marker-gene correlations.
• The framework significantly improves computational efficiency, reducing runtime and memory usage.

Conclusions:

• scMGCL provides a powerful and efficient tool for integrating single-cell ATAC-seq and RNA-seq data.
• The method facilitates deeper exploration of cell-type similarity and functional consistency.
• This framework advances the field of single-cell multi-omics data analysis.