Written and maintained by CASRAI Editorial Board
Last updated
CELLxGENE is a suite of tools from the Chan Zuckerberg Initiative for finding, downloading, exploring, analyzing, annotating and publishing single-cell datasets. Its public data portal is CZ CELLxGENE Discover. This guide maps the parts of the platform, explains how a researcher uses them, and notes what to record when citing data taken from it. It is written for researchers, data managers and research administrators who need to know what the platform is before relying on it.
What CELLxGENE is
CELLxGENE is a collection of tools, not a single application. The official documentation describes it as “a suite of tools that help scientists to find, download, explore, analyze, annotate, and publish single cell datasets.” The home page presents several entry points, summarised below.
| Component | What it does (per the official site) | Typical user |
|---|---|---|
| Collections and Datasets | Browse and download published single-cell datasets | Anyone looking for data |
| Explorer | Interactive, no-code analysis of a single dataset in the browser | Bench scientists, reviewers |
| Gene Expression | Gene expression tool listed on the home page | Researchers checking a gene of interest |
| Differential Expression | Differential expression tool listed on the home page | Researchers comparing cell populations |
| Cell Guide | An encyclopedia of cell types | Anyone needing a cell-type reference |
| Annotate | Annotation of your own data | Data submitters |
| Census | A publicly hosted data object and API for querying all single-cell RNA data from Discover | Computational biologists |
Who runs it and what it covers
The platform is operated by the Chan Zuckerberg Initiative (CZI). Discover is described on the official site as a platform for exploring and downloading single-cell data from human and mouse tissues. The Census extends this to a data object covering all single-cell RNA data in Discover, and its documentation lists Homo sapiens and Mus musculus, with the 2025-11-08 long-term-support release also including Macaca mulatta, Callithrix jacchus and Pan troglodytes.
Size figures (cells, datasets, cell types) are shown in several places on the site and are updated as data are added, so check the live home page rather than relying on a number quoted elsewhere.
How to find and explore data
The documentation is organised into sections: Find Published Data, Contribute and Publish Data, Download Published Data, Analyze Public Data, Annotate and Analyze Your Data, and a FAQ. The workflow it describes is short:
- Open CZ CELLxGENE Discover and use the Collections and Datasets pages to find a dataset (see “Find Published Data” in the docs).
- Open the dataset’s entry and read its description.
- Use the explore button below the description to open the dataset in CELLxGENE Explorer.
- For files, use the options described under “Download Published Data.”
- For programmatic work, use Census, which provides Python and R APIs; slices can be loaded as AnnData objects (Python) or Seurat and SingleCellExperiment objects (R).
Sign-in is only relevant if you are contributing data; see the “Contribute and Publish Data” section of the documentation.
A caveat when counting cells: duplicates across datasets
The Census documentation states that some cells are duplicated across multiple datasets. The cell metadata variable is_primary_data lets you include or exclude those duplicates. Anyone reporting cell counts or building a training set from Census should decide deliberately which setting to use and report it, because totals differ depending on that choice.
How to cite it
The documentation includes a “Cite cellxgene in your publications” section. The Census documentation says to follow the CZ CELLxGENE Discover citation guidelines when citing the project, and to see its tutorial “Generating citations for Census slices” for citing individual studies. A record that supports reproducibility should therefore name the platform, the Census release or dataset version used, and the original study for each dataset. This pairs naturally with a data management plan that names where reused data came from.
Strengths and limits
- Strength: a no-code route (Explorer) and a code route (Census) over the same corpus, so a reviewer and a computational analyst can look at the same data.
- Strength: Census is built on the SOMA specification with TileDB-SOMA, so you can query a slice without downloading everything.
- Limit: scope is single-cell data published through Discover, mainly human and mouse; it is not a general repository for every omics data type.
- Limit: duplicate cells across datasets (above) need an explicit decision.
- Limit: the documentation gives short guidance on the interactive workflow; deeper analysis is expected to happen in your own tools. For background on that stage see single-cell RNA sequencing workflow, Leiden clustering and CellTypist.
Alternatives
| Resource | What its own site says |
|---|---|
| CZ CELLxGENE Discover | Platform for exploring and downloading single-cell data from human and mouse tissues |
| HCA Data Portal | Portal to “explore the datasets of the Human Cell Atlas”; describes its data as “Community generated, multi-omic, open data” and offers Explore Data and Contribute Data options (read 2026-11-08). See the Human Cell Atlas entry. |
The two are different entry points and may index overlapping studies; check each portal’s own terms before reuse.
FAQ
Is CELLxGENE the same as the Census?
No. Census is a hosted data object and API for querying all single-cell RNA data from Discover; Explorer and Discover are the browser tools.
Can I use it without coding?
Yes. Explorer provides interactive, no-code analysis of a single dataset.
How do I publish my own data on it?
The documentation has a “Contribute and Publish Data” section, opened from its question “Interested in publishing your data on CELLxGENE Discover?”
Where can I read more about single-cell methods?
See multi-omics and Perturb-seq for related methods.
Sources
- CELLxGENE documentation, https://cellxgene.cziscience.com/docs (read 2026-11-08)
- CZ CELLxGENE Discover home page, https://cellxgene.cziscience.com/ (read 2026-11-08)
- CZ CELLxGENE Discover Census documentation, https://chanzuckerberg.github.io/cellxgene-census/ (read 2026-11-08)
- Human Cell Atlas Data Portal, https://data.humancellatlas.org/ (read 2026-11-08)
