UCE
Stanford SNAPUCE is the Stanford SNAP lab's universal cell embedding model, a self-supervised transformer that maps single-cell transcriptomes from many tissues and species into one latent space. It ships 4-layer and 33-layer pretrained checkpoints and an evaluation script that embeds an AnnData dataset directly, with no labeling, training, or fine-tuning step.
The repository is an inference implementation; the pretraining corpus and the model's behavior are documented in the Nature paper rather than in the README. Verified 2026-09-10 via the GitHub API, the repository README and LICENSE, and the published paper.
Openness
3 high confidence- weights
- open(ungated 4-layer and 33-layer checkpoints, fetched by the evaluation script or from Figshare)
- data
- documented-not-released(the paper describes the 36-million-cell Integrated Mega-scale Atlas corpus while the repository links processed paper datasets rather than the pretraining corpus)
- code
- partial(embedding and evaluation script only
- license
- CC-BY-4.0(governs the Figshare-hosted checkpoints
The checkpoints are freely downloadable and the corpus is described in the paper, but the repository ships only the code needed to embed a dataset, so the full-pipeline rungs are out of reach. The licence recorded here is the weights licence, not the repository's: the 33-layer checkpoint ships from Figshare item 24320806, whose record states CC BY 4.0, while the repository LICENSE is MIT and covers the code. Under multi_sku_rule the distributed weights govern, which resolves the tier to permissive_non_osi rather than osi. The score is unchanged - the walk reaches the fallthrough either way.
- https://raw.githubusercontent.com/snap-stanford/UCE/main/README.md recorded 2026-09-10
Documents automatic download of the pretrained 4-layer model and a Figshare link for the 33-layer model, exposes only eval_single_anndata.py, and links processed paper datasets rather than the training corpus.
- https://www.nature.com/articles/s41586-026-10689-z recorded 2026-09-10
Describes self-supervised training on a large corpus of cell data and the resulting Integrated Mega-scale Atlas of 36 million cells from hundreds of experiments, dozens of tissues, and eight species.
- https://raw.githubusercontent.com/snap-stanford/UCE/main/LICENSE recorded 2026-09-10
MIT License body covering the repository and the released model files, read directly rather than from the license endpoint.
Adoption
1 medium confidenceNo Hugging Face or package artifact is declared, so 338 GitHub stars are the last-resort signal; checkpoints are served from the authors' own hosting and from Figshare, neither of which reports a download count.
- https://api.github.com/repos/snap-stanford/UCE recorded 2026-09-10
stargazers_count: 338; archived: false; fork: false.
Capability
1 high confidenceThe absent adaptation pathway is a positive claim by the publisher rather than a gap in the evidence. The model emits embeddings rather than generated outputs, takes one transcriptomic modality, and operates at one scale; embedding cells from many species is coverage, not a second instrument.
- https://www.nature.com/articles/s41586-026-10689-z recorded 2026-09-10
States that UCE's universality means new cells can be embedded with no data labelling, model training or fine-tuning, and describes the output as a unified biological latent space.
- https://raw.githubusercontent.com/snap-stanford/UCE/main/README.md recorded 2026-09-10
Exposes a single embedding script whose output is UCE embeddings added to the input AnnData, with no training, fine-tuning, or generation entry point.
Verified 2026-09-10