CellAnnotator computes marker genes per cluster and queries OpenAI models for corresponding annotations. Optionally, you can provide a list of expected cell types and their markers to anchor the model. We include an additional step to harmonize annotations across samples of more complex datasets.
Cell type annotations derived with CellAnnotator can serve as an initial step in a labeling pipeline. For example, coarse labels can be used to guide semi-supervised models like scPoli or scANVI, or they can be used to compare different integration strategies with scIB metrics.
Excited to see moscot out in Nature! This was a great collaborative effort, thanks to everyone involved for making this project a success!
Happy to share CellAnnotator, a lightweight Python package to query OpenAI models for initial cell type annotations: github.com/quadbio/cell...
Inspired by ideas in www.nature.com/articles/s41... and github.com/VPetukhov/GP..., implemented as an scVerse ecosystem package with docs and tutorials.
On my first day of joining bluesky, our CellRank project is officially awarded the Helmholtz Software award for scientific originality! This is a great honor for the CellRank dev team, including Philipp Weiler, Michal Klein, and myself. Check out cellrank.org to learn more.