anndata.read_h5ad(filename, backed=None, *, as_sparse=(), as_sparse_fmt=<class 'scipy.sparse._csr.csr_matrix'>, chunk_size=6000)[source]#

Read .h5ad-formatted hdf5 file.

filename str | Path

File name of data file.

backed Union[Literal['r', 'r+'], bool, None] (default: None)

If 'r', load AnnData in backed mode instead of fully loading it into memory (memory mode). If you want to modify backed attributes of the AnnData object, you need to choose 'r+'.

Currently, backed only support updates to X. That means any changes to other slots like obs will not be written to disk in backed mode. If you would like save changes made to these slots of a backed AnnData, write them to a new file (see write()). For an example, see [here] (

as_sparse Sequence[str] (default: ())

If an array was saved as dense, passing its name here will read it as a sparse_matrix, by chunk of size chunk_size.

as_sparse_fmt type[spmatrix] (default: <class 'scipy.sparse._csr.csr_matrix'>)

Sparse format class to read elements from as_sparse in as.

chunk_size int (default: 6000)

Used only when loading sparse dataset that is stored as dense. Loading iterates through chunks of the dataset of this row size until it reads the whole dataset. Higher size means higher memory consumption and higher (to a point) loading speed.

Return type: