anndata

You can store annotated single-cell data matrices by combining the data matrix X with rich per-observation and per-variable annotations using AnnData. It supports sparse matrices for large datasets and manages obs, var, layers, embeddings, and unstructured annotations for scalable analysis.

Reading and writing h5ad files is supported natively for single-cell genomics workflows. You can load an .h5ad file, inspect metadata, and perform subsetting to validate the workflow, with interoperability also available for zarr formats across the broader scverse ecosystem.

Handling large scRNA-seq datasets efficiently requires memory-efficient sparse data handling and on-disk backed storage support. AnnData provides compact data containers with sparse matrix storage to process large annotated matrices while maintaining reproducible analysis pipelines.

Yes, AnnData integrates directly with Scanpy, Muon, and the scverse ecosystem for preprocessing, integration, batch correction, and downstream analyses. It provides the underlying data container supporting common workflows across multiple experiments.

Subsetting and concatenation across multiple single-cell experiments are supported through robust APIs. You can subset annotated data matrices and concatenate multiple experiments while preserving obs, var, layers, and unstructured metadata for reproducible pipelines.

You need AnnData for single-cell omics data management when combining a data matrix with rich per-observation and per-variable annotations and multiple representations. It is applicable across preprocessing, analysis, and interoperability workflows requiring scalable and reproducible data containers.