CELLxGENE Discover Census: Standardized Human and Mouse Single-Cell RNA Data

The CELLxGENE Discover Census provides standardized single-cell RNA data from human and mouse tissues, accessible via a public API. It aggregates data from hundreds of source datasets into a count matrix of tens of millions of cells by over 60,000 genes, accompanied by cell and gene metadata. The data is hosted by the Chan Zuckerberg Biohub and is freely available under a CC-BY-4.0 license.

Use Cases

• Train machine learning models on single-cell data based on the tens of millions of cells available.
• Analyze gene expression patterns across tissues based on standardized cell metadata like cell type and tissue.
• Filter and query specific cell populations based on metadata variables such as donor ID or sequencing technology.
• Benchmark computational methods using the standardized, multi-dataset corpus.

Strengths

• Aggregates data from hundreds of source datasets, representing the largest standardized collection of its kind.
• Includes tens of millions of cells and over 60,000 genes in its count matrix.
• Data is fully standardized and hosted for free public access with a CC-BY-4.0 license.
• API enables efficient querying and data can be transformed to common analysis objects like NumPy, Pandas, and Seurat.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for specific scale requirements.
• Last update date is unknown; freshness unverified.

Provenance

Source

Chan Zuckerberg Biohub

Collection Method

Aggregated and standardized from hundreds of individual single-cell studies.