spaDWGCN: a spatial distance weighted graph convolution network for accurate spatial tissue structure identification.

spaDWGCN improves spatial tissue structure identification in spatial transcriptomics by using distance-weighted graphs and microenvironment features across MERFISH, STARmap, Stereo-seq, Slide-seqV2, and 10× Visium.

• Why it matters: Many spatial architecture methods use binary neighbor graphs that treat all nearby cells equally, and they rarely integrate distance information into both graph construction and feature representation.
• What they did: spaDWGCN builds a distance-weighted adjacency matrix and a distance-weighted microenvironment composition matrix, then uses separate graph convolution branches for gene expression and microenvironment inputs with trainable fusion.
• The result: On multi-platform spatial transcriptomics datasets (MERFISH, STARmap, Stereo-seq, Slide-seqV2, and 10× Visium), the distance-weighted strategy achieves competitive performance for spatial domain identification and related downstream analyses.

AI summary of the abstract · may contain errors · not medical advice · read the paper.