Subcellular spatial transcriptomics resolves RNA where tissue structures extend beyond nuclei, but current aggregation strategies either impose fixed areal units or use segmented cells as anchors. This leaves process-rich and extracellular compartments difficult to measure directly. Here we report Kintsugi, a deterministic tessellation method that partitions sparse count matrices by captured-UMI density and assigns every in-tissue bin. It separates Pearson-residual gene composition from captured transcript density, preserving complete tissue representation without histology or nuclear segmentation. In mouse brain Visium HD, Kintsugi recovered nucleus-poor neuropil compartments enriched for glial and dendritic transcripts, and Xenium molecule coordinates supported dendritic mRNA localisation away from nuclei. The same representation identified nucleus-poor fibrotic scar in idiopathic pulmonary fibrosis and matrix structure in fetal cartilage. CODEX proteomics further showed that captured density was not reducible to nuclear packing alone. These results show that nucleus-poor tissue spaces contain structured, interpretable RNA biology that becomes accessible through complete tissue tessellation.
Yang, C., Zhang, X., Chen, J.
Advertisement
Stats
- Recommendations n/a n/a positive of 0 vote(s)
- Views 10
- Comments 0