Multi-omics integration for both single-cell and spatially resolved data based on dual-path graph attention...

Area of Science:

• Computational biology
• Genomics
• Proteomics

Background:

• Single-cell multi-omics integration provides high-resolution biological insights.
• Spatial multi-omics integration reveals cellular heterogeneity and spatial relationships.
• Current methods often lack spatial awareness or require improvement for integrated analysis.

Purpose of the Study:

• To develop a robust multi-omics integration method applicable to both single-cell and spatially resolved data.
• To address limitations in existing methods by incorporating spatial information.
• To facilitate a more comprehensive understanding of complex biological systems.

Main Methods:

• Proposed a dual-path graph attention auto-encoder (SSGATE) for multi-omics integration.
• SSGATE constructs neighborhood graphs using single-cell expression profiles or spatial coordinates.
• Employs self-supervised learning via graph attention auto-encoders for data integration.

Main Results:

• SSGATE successfully integrated transcriptomics and proteomics data from various tissues.
• The method was applied to both single-cell and spatially resolved datasets.
• Demonstrated superior performance and robustness compared to existing methods.

Conclusions:

• SSGATE offers a versatile solution for multi-omics data integration, accommodating both single-cell and spatial contexts.
• The method enhances downstream analyses by effectively leveraging combined molecular and spatial information.
• This advancement supports more accurate and comprehensive biological system studies.