Cell signaling characterization for spatial transcriptomics (ST) data using network analysis

Area of Science:

• Computational Biology
• Spatial Transcriptomics
• Systems Biology

Background:

• Spatial Transcriptomics (ST) enables gene expression analysis within tissue context.
• Understanding cell-cell communication is crucial for interpreting tissue architecture and function.
• Existing methods may not fully capture the spatial dynamics of ligand-receptor interactions.

Purpose of the Study:

• To introduce a novel network analysis-based method for quantifying cell-cell communication in ST data.
• To model ligand-receptor interactions using a weighted, directed network approach.
• To validate the method's ability to capture spatial signaling heterogeneity.

Main Methods:

• Constructed a network model where nodes are ST locations and edge weights reflect ligand-receptor expression and spatial distance.
• Utilized weighted in-degree centrality to quantify signaling activity for specific interactions.
• Validated the method on a real ST dataset comparing it with five existing strategies.

Main Results:

• The method successfully captures simultaneous expression heterogeneity of ligands and receptors.
• Generated biologically plausible cell communication profiles for Wnt3-Fzd1, Ephb1-Efnb3, and Ptprc-Cd22 interactions.
• Demonstrated the importance of using low-dimensional embeddings for network modeling in ST data.

Conclusions:

• The network analysis approach provides a robust framework for inferring cell-cell communication from ST data.
• The method effectively models spatial signaling, considering both expression levels and physical proximity.
• Low-dimensional gene embeddings are critical for building accurate network models in spatial transcriptomics.