cellxgene-census

You can query single-cell transcriptomics metadata at scale by using context-managed Census access with value filters. This approach lets you inspect cell types, tissues, and diseases across versioned Census releases without downloading the full dataset locally.

The best way to extract gene expression slices is by querying organism-specific experiments with out-of-core axis queries. This retrieves focused subsets and summary counts for specific tissues like human lung T cells without handling the full dataset manually.

Yes, the CELLxGENE Census supports spatial transcriptomics data queries. You can retrieve spatial slices and compare reference atlases across organisms using optional spatial workflows integrated into the versioned public data releases.

Yes, you can integrate Census query results with Scanpy or PyTorch pipelines. The retrieved gene-expression subsets and embeddings can be exported for targeted analyses and downstream reproducible analysis workflows.

No, you do not need TileDB-SOMA-ML workflows to retrieve source H5AD URIs. You can retrieve URIs and summary counts using standard out-of-core axis queries, while TileDB-SOMA-ML remains an optional workflow for specialized machine learning integration.

Context-managed Census access ensures reproducible analysis when running population-scale cell metadata searches. It safely handles versioned public single-cell data, applying value filters to query specific organisms, assays, and cell types without resource conflicts.