cellxgene-census

To query single-cell gene expression data, you filter cells by metadata like cell type, tissue, and disease to retrieve specific expression matrices from large-scale datasets. This enables precise extraction of targeted genomic profiles.

Yes, you can train PyTorch machine learning models on large-scale single-cell genomics data by integrating the queried expression matrices directly into your pipeline. This supports scalable model training across millions of cells.

Yes, the cellxgene-census skill works with scanpy, allowing you to integrate queried gene expression matrices directly into standard single-cell analysis workflows. This streamlines downstream data processing and visualization.

The best way to analyze population-scale single-cell data out-of-core is using tiledbsoma and pyarrow, which support scalable processing without loading the entire dataset into memory. This handles massive genomic datasets efficiently.

You filter cells by disease status by applying metadata queries to the cellxgene-census dataset, retrieving only the gene expression profiles matching your specified disease conditions. This narrows the dataset for targeted analysis.

A limitation when retrieving millions of cells is memory capacity, but the skill supports out-of-core processing via tiledbsoma to handle massive datasets. This prevents memory overflow during large-scale genomic queries.