Inferring spatial single-cell-level interactions through interpreting cell state and niche correlations learned by...

Area of Science:

• Computational Biology
• Genomics
• Systems Biology

Background:

• Cell-cell interactions (CCI) are crucial for tissue development and function.
• Spatial transcriptomics enables single-cell resolution analysis of CCI, but faces challenges in data interpretation and spatial encoding.

Purpose of the Study:

• To introduce GITIII, a self-supervised graph transformer model for inferring and interpreting cell-cell interactions from spatial transcriptomics data.
• To address limitations in current CCI analysis methods, including limited ligand-receptor pairs and interpretability.

Main Methods:

• Developed GITIII, a lightweight, interpretable, self-supervised graph transformer model.
• Conceptualized cells as words and their neighborhood as context to infer CCI.
• Analyzed gene expression correlation between cells and their microenvironment (niche).

Main Results:

• GITIII effectively identified and statistically interpreted spatial CCI patterns across diverse datasets.
• Demonstrated ability to understand sender cell influence on receiver cell gene expression.
• Enabled visualization of spatial CCI patterns, CCI-informed cell clustering, and network construction.

Conclusions:

• GITIII provides a powerful and interpretable framework for analyzing cell-cell interactions using spatial transcriptomics.
• The model successfully uncovered CCI in brain and tumor microenvironments, advancing biological understanding.