anndata 0.12.5__tar.gz → 0.12.6__tar.gz

{anndata-0.12.5 → anndata-0.12.6}/.github/workflows/test-gpu.yml

@@ -66,7 +66,8 @@ jobs:
         uses: astral-sh/setup-uv@v6 # TODO: upgrade once cirun image supports node 24
         with:
           enable-cache: true
-          python-version: ${{ env.max_python_version }}
+          # Any Cuda 14+ will support Python 3.14: https://github.com/cupy/cupy/issues/9346
+          python-version: '3.13'  # ${{ env.max_python_version }}
 
       - name: Install AnnData
         run: |

{anndata-0.12.5 → anndata-0.12.6}/PKG-INFO

@@ -1,12 +1,12 @@
 Metadata-Version: 2.4
 Name: anndata
-Version: 0.12.5
+Version: 0.12.6
 Summary: Annotated data.
 Project-URL: Documentation, https://anndata.readthedocs.io/
 Project-URL: Source, https://github.com/scverse/anndata
 Project-URL: Home-page, https://github.com/scverse/anndata
-Author: Philipp Angerer, Alex Wolf, Isaac Virshup, Sergei Rybakov
-Maintainer-email: Isaac Virshup <ivirshup@gmail.com>, Philipp Angerer <philipp.angerer@helmholtz-munich.de>, Ilan Gold <ilan.gold@helmholtz-munich.de>
+Author: Philipp Angerer, Alex Wolf, Isaac Virshup, Sergei Rybakov, Ilan Gold
+Maintainer-email: Philipp Angerer <philipp.angerer@helmholtz-munich.de>, Ilan Gold <ilan.gold@helmholtz-munich.de>
 License-Expression: BSD-3-Clause
 License-File: LICENSE
 Classifier: Environment :: Console
@@ -21,6 +21,7 @@ Classifier: Programming Language :: Python :: 3
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
 Classifier: Topic :: Scientific/Engineering :: Bio-Informatics
 Classifier: Topic :: Scientific/Engineering :: Visualization
 Requires-Python: >=3.11
@@ -78,7 +79,7 @@ Requires-Dist: joblib; extra == 'test'
 Requires-Dist: loompy>=3.0.5; extra == 'test'
 Requires-Dist: matplotlib; extra == 'test'
 Requires-Dist: openpyxl; extra == 'test'
-Requires-Dist: pyarrow<21; extra == 'test'
+Requires-Dist: pyarrow; extra == 'test'
 Requires-Dist: pytest-cov; extra == 'test'
 Requires-Dist: pytest-memray; extra == 'test'
 Requires-Dist: pytest-mock; extra == 'test'
@@ -100,7 +101,7 @@ Requires-Dist: joblib; extra == 'test-min'
 Requires-Dist: loompy>=3.0.5; extra == 'test-min'
 Requires-Dist: matplotlib; extra == 'test-min'
 Requires-Dist: openpyxl; extra == 'test-min'
-Requires-Dist: pyarrow<21; extra == 'test-min'
+Requires-Dist: pyarrow; extra == 'test-min'
 Requires-Dist: pytest-cov; extra == 'test-min'
 Requires-Dist: pytest-memray; extra == 'test-min'
 Requires-Dist: pytest-mock; extra == 'test-min'

anndata-0.12.6/ci/min-constraints.txt

@@ -0,0 +1 @@
+pyarrow<21

{anndata-0.12.5 → anndata-0.12.6}/docs/conf.py

@@ -134,6 +134,7 @@ intersphinx_mapping = dict(
     obstore=("https://developmentseed.org/obstore/latest/", None),
     pandas=("https://pandas.pydata.org/pandas-docs/stable", None),
     # TODO: switch to `/3` once docs are built with Python 3.14
+    # https://github.com/readthedocs/readthedocs.org/issues/12523
     python=("https://docs.python.org/3.13", None),
     scipy=("https://docs.scipy.org/doc/scipy", None),
     sklearn=("https://scikit-learn.org/stable", None),

anndata-0.12.6/docs/release-notes/0.12.6.md

@@ -0,0 +1,6 @@
+(v0.12.6)=
+### 0.12.6 {small}`2025-11-06`
+
+#### Bug fixes
+
+- Attach {class}`h5py.File` object to {class}`~anndata.AnnData` at the `file` attribute returned from  {func}`~anndata.experimental.read_lazy` {user}`ilan-gold` ({pr}`2204`)

{anndata-0.12.5 → anndata-0.12.6}/hatch.toml

@@ -21,7 +21,7 @@ env-vars.UV_CONSTRAINT = "ci/constraints.txt"
 overrides.matrix.deps.env-vars = [
     { if = [ "pre" ], key = "UV_PRERELEASE", value = "allow" },
     { if = [ "pre" ], key = "UV_CONSTRAINT", value = "ci/pre-deps.txt" },
-    { if = [ "min" ], key = "UV_CONSTRAINT", value = "ci/constraints.txt ci/min-deps.txt" },
+    { if = [ "min" ], key = "UV_CONSTRAINT", value = "ci/constraints.txt ci/min-constraints.txt ci/min-deps.txt" },
 ]
 overrides.matrix.deps.pre-install-commands = [
     { if = [
@@ -35,7 +35,10 @@ overrides.matrix.deps.pre-install-commands = [
 ]
 overrides.matrix.deps.python = [
     { if = [ "min" ], value = "3.11" },
-    { if = [ "stable", "pre" ], value = "3.13" },
+    # transitive test dep numba doesn’t support 3.14 in a stable release yet:
+    # https://github.com/numba/numba/issues/9957
+    { if = [ "stable" ], value = "3.13" },
+    { if = [ "pre" ], value = "3.14" },
 ]
 overrides.matrix.deps.features = [
     { if = [ "stable", "pre" ], value = "test" },

{anndata-0.12.5 → anndata-0.12.6}/pyproject.toml

@@ -12,9 +12,9 @@ authors = [
     { name = "Alex Wolf" },
     { name = "Isaac Virshup" },
     { name = "Sergei Rybakov" },
+    { name = "Ilan Gold" },
 ]
 maintainers = [
-    { name = "Isaac Virshup", email = "ivirshup@gmail.com" },
     { name = "Philipp Angerer", email = "philipp.angerer@helmholtz-munich.de" },
     { name = "Ilan Gold", email = "ilan.gold@helmholtz-munich.de" },
 ]
@@ -32,6 +32,7 @@ classifiers = [
     "Programming Language :: Python :: 3.11",
     "Programming Language :: Python :: 3.12",
     "Programming Language :: Python :: 3.13",
+    "Programming Language :: Python :: 3.14",
     "Topic :: Scientific/Engineering :: Bio-Informatics",
     "Topic :: Scientific/Engineering :: Visualization",
 ]
@@ -96,7 +97,7 @@ test-min = [
     "httpx<1.0",         # For data downloading
     "dask[distributed]",
     "awkward>=2.3.2",
-    "pyarrow<21",        # https://github.com/scikit-hep/awkward/issues/3579
+    "pyarrow",
     "anndata[dask]",
 ]
 test = [ "anndata[test-min,lazy]" ]

{anndata-0.12.5 → anndata-0.12.6}/src/anndata/_core/anndata.py

@@ -964,7 +964,11 @@ class AnnData(metaclass=utils.DeprecationMixinMeta):  # noqa: PLW1641
     @property
     def isbacked(self) -> bool:
         """`True` if object is backed on disk, `False` otherwise."""
-        return self.filename is not None
+        is_filename_none = self.filename is not None
+        is_x_none = (
+            getattr(self._adata_ref if self._is_view else self, "_X", None) is None
+        )
+        return is_filename_none and is_x_none
 
     @property
     def is_view(self) -> bool:
@@ -1418,7 +1422,7 @@ class AnnData(metaclass=utils.DeprecationMixinMeta):  # noqa: PLW1641
 
     @old_positionals("copy")
     def to_memory(self, *, copy: bool = False) -> AnnData:
-        """Return a new AnnData object with all backed arrays loaded into memory.
+        """Return a new AnnData object with all non-in-memory arrays loaded into memory.
 
         Params
         ------

{anndata-0.12.5 → anndata-0.12.6}/src/anndata/_core/file_backing.py

@@ -27,15 +27,24 @@ class AnnDataFileManager:
     def __init__(
         self,
         adata: anndata.AnnData,
-        filename: PathLike[str] | str | None = None,
-        filemode: Literal["r", "r+"] | None = None,
+        file_name: PathLike[str] | str | None = None,
+        file_mode: Literal["r", "r+"] | None = None,
+        file_obj: h5py.File | None = None,
     ):
+        if file_obj is not None and (file_name is not None or file_mode is not None):
+            msg = "Cannot provide both a h5py.File and the name and/or mode arguments to constructor"
+            raise ValueError(msg)
         self._adata_ref = weakref.ref(adata)
-        self.filename = filename
-        self._filemode = filemode
-        self._file = None
-        if filename:
-            self.open()
+        if file_obj is not None:
+            self.filename = filename(file_obj)
+            self._filemode = file_obj.mode
+            self._file = file_obj
+        else:
+            self.filename = file_name
+            self._filemode = file_mode
+            self._file = file_obj
+            if file_name and not self._file:
+                self.open()
 
     def __getstate__(self):
         state = self.__dict__.copy()
@@ -82,16 +91,16 @@ class AnnDataFileManager:
         return self._filename
 
     @filename.setter
-    def filename(self, filename: PathLike[str] | str | None):
-        self._filename = None if filename is None else Path(filename)
+    def filename(self, file_name: PathLike[str] | str | None):
+        self._filename = None if file_name is None else Path(file_name)
 
     def open(
         self,
-        filename: PathLike[str] | str | None = None,
+        file_name: PathLike[str] | str | None = None,
         filemode: Literal["r", "r+"] | None = None,
     ):
-        if filename is not None:
-            self.filename = filename
+        if file_name is not None:
+            self.filename = file_name
         if filemode is not None:
             self._filemode = filemode
         if self.filename is None:

{anndata-0.12.5 → anndata-0.12.6}/src/anndata/experimental/backed/_io.py

@@ -8,6 +8,7 @@ from typing import TYPE_CHECKING
 
 import h5py
 
+from anndata._core.file_backing import AnnDataFileManager
 from anndata._io.specs.registry import read_elem_lazy
 from anndata._types import AnnDataElem
 from testing.anndata._doctest import doctest_needs
@@ -28,7 +29,7 @@ if TYPE_CHECKING:
 @doctest_needs("xarray")
 @requires_xarray
 def read_lazy(
-    store: PathLike[str] | str | MutableMapping | ZarrGroup | h5py.Dataset,
+    store: PathLike[str] | str | MutableMapping | ZarrGroup | h5py.File | h5py.Group,
     *,
     load_annotation_index: bool = True,
 ) -> AnnData:
@@ -40,6 +41,9 @@ def read_lazy(
     ----------
     store
         A store-like object to be read in.  If :class:`zarr.Group`, it is best for it to be consolidated.
+        If a path to an ``.h5ad`` file is provided, the open HDF5 file will be attached to the {class}`~anndata.AnnData` at the `file` attribute and it will be the user’s responsibility to close it when done with the returned object.
+        For this reason, it is recommended to use an {class}`h5py.File` as the `store` argument when working with h5 files.
+        It must remain open for at least as long as this returned object is in use.
     load_annotation_index
         Whether or not to use a range index for the `{obs,var}` :class:`xarray.Dataset` so as not to load the index into memory.
         If `False`, the real `index` will be inserted as `{obs,var}_names` in the object but not be one of the `coords` thereby preventing read operations.
@@ -83,10 +87,11 @@ def read_lazy(
     AnnData object with n_obs × n_vars = 490 × 33452
         obs: 'donor_id', 'self_reported_ethnicity_ontology_term_id', 'organism_ontology_term_id'...
     """
-    is_h5_store = isinstance(store, h5py.Dataset | h5py.File | h5py.Group)
-    is_h5 = (
+    is_store_arg_h5_store = isinstance(store, h5py.Dataset | h5py.File | h5py.Group)
+    is_store_arg_h5_path = (
         isinstance(store, PathLike | str) and Path(store).suffix == ".h5ad"
-    ) or is_h5_store
+    )
+    is_h5 = is_store_arg_h5_path or is_store_arg_h5_store
 
     has_keys = True  # true if consolidated or h5ad
     if not is_h5:
@@ -104,7 +109,7 @@ def read_lazy(
                 f = zarr.open_group(store, mode="r")
         else:
             f = store
-    elif is_h5_store:
+    elif is_store_arg_h5_store:
         f = store
     else:
         f = h5py.File(store, mode="r")
@@ -151,6 +156,7 @@ def read_lazy(
         return func(elem)
 
     with settings.override(check_uniqueness=load_annotation_index):
-        adata = read_dispatched(f, callback=callback)
-
+        adata: AnnData = read_dispatched(f, callback=callback)
+    if is_store_arg_h5_path and not is_store_arg_h5_store:
+        adata.file = AnnDataFileManager(adata, file_obj=f)
     return adata

{anndata-0.12.5 → anndata-0.12.6}/src/anndata/tests/helpers.py

@@ -630,8 +630,9 @@ def assert_equal_arrayview(
 
 @assert_equal.register(BaseCompressedSparseDataset)
 @assert_equal.register(sparse.spmatrix)
+@assert_equal.register(CSArray)
 def assert_equal_sparse(
-    a: BaseCompressedSparseDataset | sparse.spmatrix,
+    a: BaseCompressedSparseDataset | sparse.spmatrix | CSArray,
     b: object,
     *,
     exact: bool = False,
@@ -641,13 +642,6 @@ def assert_equal_sparse(
     assert_equal(b, a, exact=exact, elem_name=elem_name)
 
 
-@assert_equal.register(CSArray)
-def assert_equal_sparse_array(
-    a: CSArray, b: object, *, exact: bool = False, elem_name: str | None = None
-):
-    return assert_equal_sparse(a, b, exact=exact, elem_name=elem_name)
-
-
 @assert_equal.register(CupySparseMatrix)
 def assert_equal_cupy_sparse(
     a: CupySparseMatrix, b: object, *, exact: bool = False, elem_name: str | None = None
@@ -880,29 +874,53 @@ def _(a):
 
 
 @singledispatch
-def as_sparse_dask_array(a) -> DaskArray:
-    import dask.array as da
-
-    return da.from_array(sparse.csr_matrix(a), chunks=_half_chunk_size(a.shape))
+def _as_sparse_dask(
+    a: NDArray | CSArray | CSMatrix | DaskArray,
+    *,
+    typ: type[CSArray | CSMatrix | CupyCSRMatrix],
+    chunks: tuple[int, ...] | None = None,
+) -> DaskArray:
+    """Convert a to a sparse dask array, preserving sparse format and container (`cs{rc}_{array,matrix}`)."""
+    raise NotImplementedError
 
 
-@as_sparse_dask_array.register(CSMatrix)
-def _(a):
+@_as_sparse_dask.register(CSArray | CSMatrix | np.ndarray)
+def _(
+    a: CSArray | CSMatrix | NDArray,
+    *,
+    typ: type[CSArray | CSMatrix | CupyCSRMatrix],
+    chunks: tuple[int, ...] | None = None,
+) -> DaskArray:
     import dask.array as da
 
-    return da.from_array(a, _half_chunk_size(a.shape))
+    chunks = _half_chunk_size(a.shape) if chunks is None else chunks
+    return da.from_array(_as_sparse_dask_inner(a, typ=typ), chunks=chunks)
 
 
-@as_sparse_dask_array.register(CSArray)
-def _(a):
-    import dask.array as da
+@_as_sparse_dask.register(DaskArray)
+def _(
+    a: DaskArray,
+    *,
+    typ: type[CSArray | CSMatrix | CupyCSRMatrix],
+    chunks: tuple[int, ...] | None = None,
+) -> DaskArray:
+    assert chunks is None  # TODO: if needed we can add a .rechunk(chunks)
+    return a.map_blocks(_as_sparse_dask_inner, typ=typ, dtype=a.dtype, meta=typ((2, 2)))
 
-    return da.from_array(sparse.csr_matrix(a), _half_chunk_size(a.shape))
 
+def _as_sparse_dask_inner(
+    a: NDArray | CSArray | CSMatrix, *, typ: type[CSArray | CSMatrix | CupyCSRMatrix]
+) -> CSArray | CSMatrix:
+    """Convert into a a sparse container that dask supports (or complain)."""
+    if issubclass(typ, CSArray):  # convert sparray to spmatrix
+        msg = "AnnData doesn't support `cs_{r,c}_array` inside Dask"
+        raise TypeError(msg)
+    if issubclass(typ, CupySparseMatrix):
+        a = as_cupy(a)  # can’t Cupy sparse constructors don’t accept numpy ndarrays
+    return typ(a)
 
-@as_sparse_dask_array.register(DaskArray)
-def _(a):
-    return a.map_blocks(sparse.csr_matrix)
+
+as_sparse_dask_matrix = partial(_as_sparse_dask, typ=sparse.csr_matrix)
 
 
 @singledispatch
@@ -947,14 +965,11 @@ except ImportError:
     format_to_memory_class = {}
 
 
-# TODO: If there are chunks which divide along columns, then a coo_matrix is returned by compute
-# We should try and fix this upstream in dask/ cupy
 @singledispatch
-def as_cupy_sparse_dask_array(a, format="csr"):
-    memory_class = format_to_memory_class[format]
-    cpu_da = as_sparse_dask_array(a)
-    return cpu_da.rechunk((cpu_da.chunks[0], -1)).map_blocks(
-        memory_class, dtype=a.dtype, meta=memory_class(cpu_da._meta)
+def as_cupy_sparse_dask_array(a, format="csr") -> DaskArray:
+    chunk_rows, _ = _half_chunk_size(a.shape)
+    return _as_sparse_dask(
+        a, typ=format_to_memory_class[format], chunks=(chunk_rows, -1)
     )
 
 
@@ -964,7 +979,8 @@ def _(a, format="csr"):
     import dask.array as da
 
     memory_class = format_to_memory_class[format]
-    return da.from_array(memory_class(a), chunks=(_half_chunk_size(a.shape)[0], -1))
+    chunk_rows, _ = _half_chunk_size(a.shape)
+    return da.from_array(memory_class(a), chunks=(chunk_rows, -1))
 
 
 @as_cupy_sparse_dask_array.register(DaskArray)
@@ -982,9 +998,9 @@ def resolve_cupy_type(val):
 
     if issubclass(input_typ, np.ndarray):
         typ = CupyArray
-    elif issubclass(input_typ, sparse.csr_matrix):
+    elif issubclass(input_typ, sparse.csr_matrix | sparse.csr_array):
         typ = CupyCSRMatrix
-    elif issubclass(input_typ, sparse.csc_matrix):
+    elif issubclass(input_typ, sparse.csc_matrix | sparse.csc_array):
         typ = CupyCSCMatrix
     else:
         msg = f"No default target type for input type {input_typ}"
@@ -1005,7 +1021,7 @@ def as_cupy(val, typ=None):
     if issubclass(typ, CupyArray):
         import cupy as cp
 
-        if isinstance(val, CSMatrix):
+        if isinstance(val, CSMatrix | CSArray):
             val = val.toarray()
         return cp.array(val)
     elif issubclass(typ, CupyCSRMatrix):
@@ -1061,7 +1077,7 @@ BASE_MATRIX_PARAMS = [
 
 DASK_MATRIX_PARAMS = [
     pytest.param(as_dense_dask_array, id="dense_dask_array"),
-    pytest.param(as_sparse_dask_array, id="sparse_dask_array"),
+    pytest.param(as_sparse_dask_matrix, id="sparse_dask_matrix"),
 ]
 
 CUPY_MATRIX_PARAMS = [

{anndata-0.12.5 → anndata-0.12.6}/tests/lazy/test_read.py

@@ -175,7 +175,7 @@ def test_view_of_view_to_memory(adata_remote: AnnData, adata_orig: AnnData):
 
 @pytest.mark.zarr_io
 def test_unconsolidated(tmp_path: Path, mtx_format):
-    adata = gen_adata((1000, 1000), mtx_format, **GEN_ADATA_NO_XARRAY_ARGS)
+    adata = gen_adata((10, 10), mtx_format, **GEN_ADATA_NO_XARRAY_ARGS)
     orig_pth = tmp_path / "orig.zarr"
     adata.write_zarr(orig_pth)
     (orig_pth / ".zmetadata").unlink()
@@ -188,6 +188,16 @@ def test_unconsolidated(tmp_path: Path, mtx_format):
     store.assert_access_count("obs/.zgroup", 1)
 
 
+def test_h5_file_obj(tmp_path: Path):
+    adata = gen_adata((10, 10), **GEN_ADATA_NO_XARRAY_ARGS)
+    orig_pth = tmp_path / "adata.h5ad"
+    adata.write_h5ad(orig_pth)
+    remote = read_lazy(orig_pth)
+    assert remote.file.is_open
+    assert remote.filename == orig_pth
+    assert_equal(remote.to_memory(), adata)
+
+
 @pytest.fixture(scope="session")
 def df_group(tmp_path_factory) -> zarr.Group:
     df = gen_typed_df(120)

{anndata-0.12.5 → anndata-0.12.6}/tests/test_dask.py

@@ -21,7 +21,7 @@ from anndata.tests.helpers import (
     as_cupy_sparse_dask_array,
     as_dense_cupy_dask_array,
     as_dense_dask_array,
-    as_sparse_dask_array,
+    as_sparse_dask_matrix,
     assert_equal,
     check_all_sharded,
     gen_adata,
@@ -288,7 +288,7 @@ def test_assign_X(adata):
     ("array_func", "mem_type"),
     [
         pytest.param(as_dense_dask_array, np.ndarray, id="dense_dask_array"),
-        pytest.param(as_sparse_dask_array, sparse.csr_matrix, id="sparse_dask_array"),
+        pytest.param(as_sparse_dask_matrix, sparse.csr_matrix, id="sparse_dask_matrix"),
         pytest.param(
             as_dense_cupy_dask_array,
             CupyArray,
@@ -326,7 +326,7 @@ def test_dask_to_memory_unbacked(array_func, mem_type):
     "array_func",
     [
         pytest.param(as_dense_dask_array, id="dense_dask_array"),
-        pytest.param(as_sparse_dask_array, id="sparse_dask_array"),
+        pytest.param(as_sparse_dask_matrix, id="sparse_dask_matrix"),
         pytest.param(
             as_dense_cupy_dask_array,
             id="cupy_dense_dask_array",

{anndata-0.12.5 → anndata-0.12.6}/tests/test_settings.py

@@ -274,8 +274,9 @@ def test_hints():
     )
     settings_types_mod = types.ModuleType(types_loader.name)
     types_loader.exec_module(settings_types_mod)
-    for settings_key in dir(settings):
-        if not settings_key.startswith("_"):
-            assert hasattr(settings_types_mod._AnnDataSettingsManager, settings_key)
-    for settings_key in dir(settings_types_mod._AnnDataSettingsManager):
-        assert hasattr(settings, settings_key)
+
+    obj_attrs, typing_attrs = (
+        {k for k in dir(o) if not k.startswith("_")}
+        for o in (settings, settings_types_mod._AnnDataSettingsManager)
+    )
+    assert obj_attrs == typing_attrs