MorphoANOVA: Dissecting Histology-Gene Expression Relationships in Spatial Transcriptomics

MorphoANOVA identifies slide-wide and location-specific morphology–expression associations for spatial transcriptomics genes, including stronger location-specific signals for invasive-cancer markers across three HER2-positive breast cancer patients.

• Why it matters: Spatial transcriptomics links gene expression to tissue architecture, but it is unclear which genes track morphology, how those links change across space, and which genes show extra spatial structure beyond morphology.
• What they did: The method uses gene-wise nested spatially varying coefficient models with sequential ANOVA to test slide-wide morphology effects, location-specific changes, and additional spatial variation after accounting for an H&E-derived feature.
• The result: Simulations show it can separate modeled components and recover spatial patterns, and applications to human cortex and breast cancer prioritize pathways and reveal distinct morphology–expression relationships tied to cell composition.

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