scGraphformer: unveiling cellular heterogeneity and interactions in scRNA-seq data using a scalable graph transformer...

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Accurate cell type classification from single-cell RNA sequencing (scRNA-seq) data is crucial for understanding cellular heterogeneity.
• Traditional graph neural network (GNN) models face limitations due to reliance on predefined graphs, hindering the exploration of intricate cell-cell interactions.

Purpose of the Study:

• To introduce scGraphformer, a transformer-based GNN designed to overcome limitations of traditional GNNs in scRNA-seq analysis.
• To develop a method that learns comprehensive cell-cell relational networks directly from scRNA-seq data.
• To enhance the identification of subtle cellular patterns and functional relationships.

Main Methods:

• Developed scGraphformer, a transformer-based GNN architecture.
• Implemented an iterative refinement process to construct a dense cell-cell relational network.
• Evaluated performance on multiple scRNA-seq datasets.

Main Results:

• scGraphformer demonstrated superior performance in cell type identification compared to existing methods.
• The model successfully captured a full spectrum of cellular interactions, revealing previously obscured patterns.
• Scalability was confirmed on large-scale datasets.

Conclusions:

• scGraphformer offers enhanced cell type classification and reveals underlying cell interactions.
• The method provides deeper insights into functional cellular relationships and cellular heterogeneity.
• scGraphformer has the potential to significantly advance single-cell analysis and understanding of cellular behavior.