Tissue module discovery in single-cell-resolution spatial transcriptomics data via cell-cell interaction-aware cell...

Area of Science:

• Computational biology
• Spatial transcriptomics
• Single-cell analysis

Background:

• Understanding tissue-specific cell functions requires analyzing spatial organization in single-cell-resolution spatial transcriptomics (ST) data.
• Existing computational methods need improvement for comprehensive ST data analysis.

Purpose of the Study:

• To develop a deep-learning method for cell-type identification and tissue module discovery from ST data.
• To capture both gene expression profiles and spatial neighbor interactions for cell representation.

Main Methods:

• Introduced SPACE (ST data analysis via interaction-aware cell embedding), a deep-learning approach.
• SPACE learns cell representations incorporating gene expression and spatial neighborhood information.
• Applied SPACE to identify spatially informed cell subtypes and discover tissue modules ('cell communities').

Main Results:

• SPACE identified spatially informed cell subtypes with distinct distribution patterns and proximal cell types.
• The method automatically discovered 'cell communities' with clear boundaries and uniform cell type distribution.
• SPACE generated characteristic proximal cell-cell interaction networks for each cell community, aiding signaling analysis.

Conclusions:

• SPACE provides a novel computational framework for analyzing single-cell-resolution ST data.
• The method enhances understanding of how proximal cell-cell interactions contribute to biological functions within tissue modules.
• SPACE facilitates large-scale ST projects aiming to decipher emergent biological functions from spatial organization.