SlotDeconv: spatial transcriptomics deconvolution via diversity-constrained prototype learning and spatial refinement
Hanzhang Fang, Cong Qi, Yuanjie Zou, Yeqing Chen, Zhi Wei · 2026
Read SlotDeconv: spatial transcriptomics deconvolution via diversity-constrained prototype learning and spatial refinement by Hanzhang Fang, Cong Qi, Yuanjie Zou, Yeqing Chen, Zhi Wei (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/PMC13501309.txt.
Description
Fang and colleagues presented SlotDeconv, a method for estimating which cell types occupy each spot in spatial transcriptomics data, in a 2026 paper. The technology measures gene activity at spots across a tissue slice. Unmixing fails when sister cell types look identical and when neighborhood information is ignored.
Fang split the problem in two: a reference module learns distinctive signature vectors for each cell type, trained with a diversity constraint that pushes similar types apart. A spatial module first estimates proportions spot by spot, then smooths them using neighborhood consistency. Removing the diversity constraint cut accuracy sharply, and using average profiles instead of learned signatures collapsed it entirely. Against four rivals on a twenty-seven-cell-type mouse brain dataset, SlotDeconv achieved the highest correlations, with the biggest gains exactly on the confused cortical neuron subtypes. It validated on human pancreatic cancer and mouse olfactory bulb tissues.
Catalog record
| CREATOR | Hanzhang Fang, Cong Qi, Yuanjie Zou, Yeqing Chen, Zhi Wei |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 28 KB |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |