Authors
D'Agostini, D., Abdelaal, T.
Abstract
Accurate discrimination between tumour and non-tumour cells is essential for interpreting spatial transcriptomics data and identifying spatially resolved therapeutic targets. Existing tumour-cell annotation approaches were primarily developed for single-cell RNA sequencing and do not explicitly incorporate the spatial organization of cells. Here, we introduce SpaceMACE, an extension of the MACE tumour-cell annotation framework for sequencing-based and imaging-based spatial transcriptomics. SpaceMACE first combines reference-based cell-type classification with copy-number-variation-based tumour classification. The resulting consensus annotations are subsequently refined using spatial proximity and gene-expression similarity. We evaluated SpaceMACE across multiple spatial transcriptomics platforms, including Stereo-seq, Visium HD, CosMx and Xenium, and across colorectal, hepatocellular and ovarian cancer datasets. SpaceMACE retained the high precision of the original MACE framework while improving the spatial coherence of tumour-cell annotations. Together, these results demonstrate that SpaceMACE provides a scalable and platform-independent framework for tumour-cell annotation in spatial transcriptomics and supports the spatial investigation of tumour-specific molecular targets.
Preprint server:
bioRxiv
The authors list and abstract were imported from bioRxiv on 18 Nov 2026.
Advertisement
Stats
- Recommendations n/a n/a positive of 0 vote(s)
- Views 4
- Comments 0