cellxgene-census

You can query CELLxGENE Census single-cell data using Python via TileDB-SOMA. This allows you to filter by cell_type, tissue, disease, and donor to retrieve obs, var, and expression matrices programmatically.

Yes, CELLxGENE Census data integrates directly with AnnData, PyTorch, and Scanpy. You can retrieve cell-level metadata and precomputed embeddings to feed into downstream machine learning and visualization workflows.

The Skill supports out-of-core processing across multiple datasets using TileDB-SOMA. This enables cross-tissue comparisons and multi-dataset analyses on large-scale single-cell data without loading it all into memory.

To ensure you only retrieve unique primary cells from the Census, you should filter your queries using the is_primary_data == True condition. This prevents duplicate cells from appearing across multiple datasets.

Yes, the Skill provides programmatic access to Census data for both Homo sapiens and Mus musculus. You can query and analyze single-cell data across these two organisms using the TileDB-SOMA API.

The Skill supports census_version control to manage which version of the Census data you are querying. This ensures reproducible single-cell data retrieval and analysis across different dataset snapshots.