adataviz.adata module

class adataviz.adata.AnnDataCollection(adata_paths: List[str], adata: AnnData | None = None, source_info: List[Dict] | None = None)

Bases: object

Wrapper representing a merged collection of AnnData files. .. rubric:: Example

ds = AnnDataCollection.from_files([“a.h5ad”, “b.h5ad”], out_path=”merged.h5ad”) view = ds[[True]*100, [“GeneA”, “GeneB”]] adata = view.to_memory()

classmethod from_files(paths: Sequence[str], out_path: str | None = None, metadata_path: str | None = None) → AnnDataCollection

Create a merged AnnDataCollection from existing .h5ad files. This will merge obs (stacked) and var (union by var_names). X is not merged; the saved on-disk AnnData will contain an empty sparse matrix of shape (n_obs_total, n_vars_total).

classmethod read(path: str) → AnnDataCollection

Load a merged .h5ad file created by from_files and return an AnnDataCollection.

Expects src_paths (or individual_adata_paths) in adata.uns and src_info which may be JSON-serialized or a list.

update_paths(mapping: Dict[str, str] | None = None, search_dirs: Sequence[str] | None = None, recursive: bool = True) → Dict[str, str | None]

Update stored source paths after files have been moved.

Parameters - mapping: optional dict mapping old_path -> new_path. If provided,

entries in adata_paths will be replaced according to this map.

• search_dirs: optional sequence of directories to search for moved files by matching basenames (first match wins). Used when mapping is not provided or doesn’t contain an entry for a given source.

• recursive: if True, search directories recursively when using search_dirs.

Returns a dict mapping the original absolute path -> new absolute path (or None if not found).

class adataviz.adata.AnnDataView(dataset: AnnDataCollection, obs_idx: ndarray | None = None, var_names: List[str] | None = None)

Bases: object

Helper representing a (cells, genes) subset of an AnnDataCollection. Call to_memory() to read and assemble the actual AnnData with X.

to_memory(thread=8) → AnnData

Load selected X chunks from underlying files and assemble AnnData. Returns a new anndata.AnnData with concatenated X for the selected cells and genes.

adataviz.adata.is_annadatacollection(path: str) → bool

Return True if path points to an AnnData file produced by AnnDataCollection.from_files. Heuristic: the AnnData must have uns entries src_paths or individual_adata_paths and its obs must contain either _source_idx or _orig_obs_name markers.

adataviz.adata.main()

adataviz.adata.test_subset()