OncoSpot enables benchmarking of cancer spatial transcriptomics methods across tumor evolutionary modes

OncoSpot simulates Visium-like spatial transcriptomics for high-grade serous ovarian carcinoma, colorectal cancer, and hepatocellular carcinoma, enabling benchmarking with known malignant fraction, subclones, and CNA profiles.

• Why it matters: Real tumor spatial transcriptomics lacks ground truth for copy-number alterations, subclonal identity, and malignant cell fraction, making method evaluation difficult and failure modes hard to detect.
• What they did: OncoSpot builds a tumor-evolution scaffold from DEMON, generates CNA-driven expression along clonal phylogenies with CNAsim, adds cancer-specific malignant programs and clustered nonmalignant cell types, then aggregates cells into spots with known labels.
• The result: In simulations, subclones were more concordant with phylogeny-space than scCube under matched populations, and benchmarking showed cancer-dependent accuracy (Clonalscope best for ovarian/colorectal; CopyKAT best for hepatocellular), including a Clonalscope malignant-spot inversion failure linked to its assumptions.

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