Reconstructing cell-cell interaction network in single-cell spatial transcriptomics via directed heterogeneous graph...

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Spatial transcriptomics provides gene expression and spatial location data, crucial for analyzing cell-cell interaction (CCI) networks.
• Existing methods for CCI inference often rely on known ligand-receptor pairs and lack directional modeling.
• Current deep learning frameworks for CCI analysis frequently use symmetric decoders or undirected graphs.

Purpose of the Study:

• To develop a novel deep learning framework, DualCellChat, for reconstructing comprehensive and accurate cell-cell interaction networks from spatial transcriptomic data.
• To address limitations of existing methods by incorporating directional information and handling incomplete single-cell spatial transcriptomics.
• To enable downstream analysis for identifying signature genes and significant ligand-receptor pairs involved in cellular communication.

Main Methods:

• Utilized a directed heterogeneous graph autoencoder architecture.
• Applied the DualCellChat approach to five single-cell spatial datasets generated using four distinct technologies.
• Developed downstream analysis modules for inferring signature genes and ligand-receptor pairs from the reconstructed CCI network.

Main Results:

• DualCellChat successfully reconstructed a complete and accurate cell-cell interaction network.
• The method demonstrated superior performance compared to existing deep learning-based approaches on benchmark datasets.
• DualCellChat inherently models the directionality of cellular interactions, providing more biologically relevant insights.
• Downstream analysis successfully identified significant ligand-receptor pairs and signature genes specific to cell types.

Conclusions:

• DualCellChat offers a powerful and accurate method for inferring directional cell-cell interaction networks from spatial transcriptomic data.
• The framework advances the analysis of cellular communication in complex biological systems.
• The approach facilitates the discovery of novel insights into cell-cell communication mechanisms and their underlying molecular players.