See every cell. Query any gene. Explore in 2D and 3D.
Cellucid is a browser-first, GPU-accelerated workspace for exploring large single-cell datasets in real time. Move naturally through 1D, 2D, and 3D embeddings, compare populations side by side, follow cellular dynamics, and turn an exploratory view into a reproducible figure or shared session.
- Fast WebGL rendering, up to 100 million cells in 3D on a 16 GB laptop
- Subsampled reading: open a seeded share of the cells (uniform, storage-aligned blocks, stratified, spatial), downloading only the parts of the file that hold them; the same cells in Python, R, links and sessions; atlases open all cells that fit, genes on a stated sample
- Interactive gene-expression and cell-metadata coloring, filtering, and highlighting
- Independent multiview comparisons with per-view colors and filter stacks
- Animated RNA-velocity and drift overlays, KNN connectivity, and volumetric smoke
- Built-in marker genes and differential expression with scanpy-compatible statistics on X, a layer or raw, dot plots, gene signatures (module score, UCell), cell-type composition across samples and conditions, correlation and quick insights
- Orbit, Planar, and Free Fly navigation with reproducible Camera Path animation
- Community annotation, voting, and consensus with optional GitHub collaboration
- An undoable, branching history of every step, replayable on another dataset, written up as methods; shareable session state, dataset links, badges, embeds, SVG and high-DPI PNG export
- AI agents act under your permissions: a chat bar with your own key, Claude or ChatGPT as a connector, or a local app via
cellucid mcp - A hover card with the values of the cell under the pointer, in every view
- Reads AnnData directly (H5AD, Zarr folders and Zarr ZIP), and Loom, 10x Cell Ranger HDF5 and
outs/folders, MuData (.h5mu, MuData Zarr), Seurat and SingleCellExperiment objects (.rds,.rda,.RData) and MatrixMarket folders as AnnData, with a Modality choice for multimodal data, from your computer, a link (CELLxGENE, GEO, Zenodo, Figshare, Hugging Face, GitHub, cloud drives or any web address), a dataset catalog, a local or remote server, or a Jupyter notebook or R session, private Drive and OneDrive files after one sign-in; the Python and R packages write Cellucid-ready AnnData
Placeholder citation. The Cellucid preprint is in preparation for bioRxiv and has no DOI yet; the reference below is a placeholder and will be replaced when the preprint is published.
If you use Cellucid, please cite:
Inecik, K. PLACEHOLDER: Cellucid preprint, title to be announced (forthcoming on bioRxiv). bioRxiv (in preparation).
BibTeX and other formats: Citing Cellucid.
Installation, data preparation, every loading path, UI controls, examples, and troubleshooting are covered in the complete Cellucid documentation.