cellxgene-census-query

You can query CZ CELLxGENE Census data programmatically by opening a Soma session and filtering by cell type, tissue, and disease to retrieve expression data and metadata as an AnnData object without downloading entire files.

Yes, querying CZ CELLxGENE Census allows you to filter datasets by cell type, tissue, and disease to retrieve expression data, metadata, and pre-calculated embeddings for targeted single-cell analyses.

Yes, querying CELLxGENE Census integrates directly with Scanpy and PyTorch, enabling you to retrieve filtered single-cell expression data and metadata for large-scale multi-dataset integration and ML pipelines.

You need Python 3.9+ and dependencies including cellxgene-census, tiledbsoma, scanpy, pyarrow, pandas, and numpy, with optional PyTorch support for experimental ML pipelines.

The best way is to query the CELLxGENE Census programmatically, utilizing out-of-core processing to integrate cross-dataset single-cell analyses and enable reproducible research without downloading entire matrices.

Filtering by dataset version, disease, tissue, and cell type works by querying the Census programmatically to retrieve specific expression data and metadata subsets, enabling targeted and reproducible single-cell analyses.