Complete spatially resolved gene expression is not necessary for identifying spatial domains.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Spatially resolved transcriptomics (SRT) is crucial for understanding tissue architecture.
• Existing SRT methods often require complex, high-resolution gene expression data.

Purpose of the Study:

• To introduce BINARY, a novel graph convolutional network for spatial domain delineation.
• To demonstrate the efficacy of using binarized SRT data for accurate spatial domain identification.

Main Methods:

• Developed a graph convolutional network (BINARY) incorporating attention and positive emphasis mechanisms.
• Utilized exclusively binarized spatially resolved transcriptomics data as input.
• Evaluated BINARY's performance across diverse SRT datasets.

Main Results:

• BINARY accurately delineates spatial domains in various SRT data types.
• Achieved superior performance compared to existing methods.
• Demonstrated effectiveness using significantly less input information (binarized data).

Conclusions:

• Precise gene expression quantification may not be essential for spatial domain delineation.
• Binarized SRT data offers a powerful and efficient alternative for spatial analysis.
• Suggests broader applications for binarized spatial gene expression data.