add squidpy backend

docs/api/scanpy_gpu.md

@@ -2,6 +2,18 @@
 
 These functions offer accelerated near drop-in replacements for common tools provided by [`scanpy`](https://scanpy.readthedocs.io/en/stable/api/index.html) {cite}`Wolf2018`.
 
+## Scanpy backend
+
+With Scanpy versions that support computational backends, RAPIDS-singlecell is available as the `rapids_singlecell` backend with the aliases `cuda`, `rapids`, `rapids-singlecell`, and `rsc`.
+
+```python
+import scanpy as sc
+
+sc.settings.backend = "cuda"
+```
+
+The backend exposes RAPIDS-singlecell's `pp` and `tl` functions, plus {func}`rapids_singlecell.get.aggregate`, for Scanpy's backend dispatcher.
+
 ## Preprocessing `pp`
 Filtering of highly-variable genes, batch-effect correction, per-cell normalization.
 

docs/api/squidpy_gpu.md

@@ -3,6 +3,18 @@
 {mod}`squidpy.gr` is a tool for the analysis of spatial molecular data {cite}`Palla2022`.
 {mod}`rapids_singlecell.gr` accelerates some of these functions.
 
+## Squidpy backend
+
+With Squidpy versions that support computational backends, RAPIDS-singlecell is available as the `rapids_singlecell` backend with the aliases `cuda`, `rapids-singlecell`, and `rsc`.
+
+```python
+import squidpy as sq
+
+sq.settings.backend = "cuda"
+```
+
+The backend exposes RAPIDS-singlecell's {mod}`rapids_singlecell.gr` functions for Squidpy's backend dispatcher.
+
 ```{eval-rst}
 .. module:: rapids_singlecell.gr
 .. currentmodule:: rapids_singlecell

docs/conf.py

@@ -79,6 +79,7 @@
     "pylibraft",
     "dask",
     "cuvs",
+    "spatialdata",
 ]
 default_role = "literal"
 napoleon_google_docstring = False
@@ -126,6 +127,7 @@
     "statsmodels": ("https://www.statsmodels.org/stable/", None),
     "omnipath": ("https://omnipath.readthedocs.io/en/latest/", None),
     "dask": ("https://docs.dask.org/en/stable/", None),
+    "spatialdata": ("https://spatialdata.scverse.org/en/stable/", None),
 }
 
 # List of patterns, relative to source directory, that match files and

docs/release-notes/index.md

@@ -3,6 +3,10 @@
 # Release notes
 
 
+## Version 0.16.0
+```{include} /release-notes/0.16.0.md
+```
+
 ## Version 0.15.0
 ```{include} /release-notes/0.15.0.md
 ```

pyproject.toml

@@ -71,6 +71,12 @@ dev = [
     "pre-commit",
 ]
 
+[project.entry-points."scanpy.backends"]
+rapids_singlecell = "rapids_singlecell._backends.scanpy"
+
+[project.entry-points."squidpy.backends"]
+rapids_singlecell = "rapids_singlecell._backends.squidpy"
+
 [project.urls]
 Documentation = "https://rapids-singlecell.readthedocs.io"
 Source = "https://github.com/scverse/rapids_singlecell"

src/rapids_singlecell/_backends/__init__.py

@@ -0,0 +1 @@
+from __future__ import annotations

src/rapids_singlecell/_backends/scanpy.py

@@ -0,0 +1,160 @@
+from __future__ import annotations
+
+from typing import TYPE_CHECKING
+
+from rapids_singlecell.get import aggregate
+from rapids_singlecell.preprocessing import (
+    bbknn,
+    calculate_qc_metrics,
+    filter_cells,
+    filter_genes,
+    filter_highly_variable,
+    flag_gene_family,
+    harmony_integrate,
+    highly_variable_genes,
+    neighbors,
+    normalize_pearson_residuals,
+    normalize_total,
+    regress_out,
+    scrublet,
+    scrublet_simulate_doublets,
+)
+from rapids_singlecell.preprocessing import log1p as _log1p
+from rapids_singlecell.preprocessing import pca as _pca
+from rapids_singlecell.preprocessing import scale as _scale
+from rapids_singlecell.preprocessing._pca import _empty
+from rapids_singlecell.tools import (
+    diffmap,
+    draw_graph,
+    embedding_density,
+    kmeans,
+    leiden,
+    louvain,
+    rank_genes_groups,
+    rank_genes_groups_logreg,
+    score_genes,
+    score_genes_cell_cycle,
+    tsne,
+    umap,
+)
+
+if TYPE_CHECKING:
+    from anndata import AnnData
+    from numpy.typing import DTypeLike, NDArray
+
+name = "rapids_singlecell"
+aliases = ["cuda", "rapids", "rapids-singlecell", "rsc"]
+
+
+def log1p(
+    data: AnnData,
+    *,
+    base: float | None = None,
+    layer: str | None = None,
+    obsm: str | None = None,
+    inplace: bool = True,
+    copy: bool = False,
+):
+    return _log1p(
+        data,
+        base=base,
+        layer=layer,
+        obsm=obsm,
+        inplace=inplace,
+        copy=copy,
+    )
+
+
+def pca(
+    data: AnnData,
+    n_comps: int | None = None,
+    *,
+    layer: str | None = None,
+    zero_center: bool = True,
+    svd_solver: str | None = None,
+    chunked: bool = False,
+    chunk_size: int | None = None,
+    rng=None,
+    mask_var: NDArray | str | None = _empty,
+    dtype: DTypeLike = "float32",
+    key_added: str | None = None,
+    copy: bool = False,
+    random_state: int | None = 0,
+    use_highly_variable: bool | None = None,
+    **kwargs,
+) -> None | AnnData:
+    if rng is not None:
+        random_state = rng
+    return _pca(
+        data,
+        n_comps=n_comps,
+        layer=layer,
+        zero_center=zero_center,
+        svd_solver=svd_solver,
+        random_state=random_state,
+        mask_var=mask_var,
+        use_highly_variable=use_highly_variable,
+        dtype=dtype,
+        chunked=chunked,
+        chunk_size=chunk_size,
+        key_added=key_added,
+        copy=copy,
+        **kwargs,
+    )
+
+
+def scale(
+    data: AnnData,
+    *,
+    zero_center: bool = True,
+    max_value: float | None = None,
+    copy: bool = False,
+    layer: str | None = None,
+    obsm: str | None = None,
+    mask_obs: NDArray | str | None = None,
+    inplace: bool = True,
+):
+    return _scale(
+        data,
+        zero_center=zero_center,
+        max_value=max_value,
+        copy=copy,
+        layer=layer,
+        obsm=obsm,
+        mask_obs=mask_obs,
+        inplace=inplace,
+    )
+
+
+__all__ = [
+    "aggregate",
+    "bbknn",
+    "calculate_qc_metrics",
+    "diffmap",
+    "draw_graph",
+    "embedding_density",
+    "filter_cells",
+    "filter_genes",
+    "filter_highly_variable",
+    "flag_gene_family",
+    "harmony_integrate",
+    "highly_variable_genes",
+    "kmeans",
+    "leiden",
+    "log1p",
+    "louvain",
+    "neighbors",
+    "normalize_pearson_residuals",
+    "normalize_total",
+    "pca",
+    "rank_genes_groups",
+    "rank_genes_groups_logreg",
+    "regress_out",
+    "scale",
+    "score_genes",
+    "score_genes_cell_cycle",
+    "scrublet",
+    "scrublet_simulate_doublets",
+    "tsne",
+    "umap",
+]

src/rapids_singlecell/_backends/squidpy.py

@@ -0,0 +1,8 @@
+from __future__ import annotations
+
+from rapids_singlecell.squidpy_gpu import co_occurrence, ligrec, spatial_autocorr
+
+name = "rapids_singlecell"
+aliases = ["rapids-singlecell", "rsc", "cuda"]
+
+__all__ = ["co_occurrence", "ligrec", "spatial_autocorr"]

src/rapids_singlecell/_compat.py

@@ -7,6 +7,11 @@
 from scipy.sparse import csc_matrix as csc_matrix_cpu
 from scipy.sparse import csr_matrix as csr_matrix_cpu
 
+try:
+    from spatialdata import SpatialData
+except ImportError:
+    SpatialData = None
+
 
 def _meta_dense(dtype):
     return cp.zeros([0], dtype=dtype)

src/rapids_singlecell/preprocessing/_normalize.py

@@ -357,7 +357,7 @@ def _calc_log1p(X: ArrayTypesDask, base: float | None = None) -> ArrayTypesDask:
 
 
 def log1p(
-    adata: AnnData,
+    data: AnnData,
     *,
     base: float | None = None,
     layer: str | None = None,
@@ -373,7 +373,7 @@ def log1p(
 
     Parameters
     ----------
-        adata
+        data
             AnnData object
         base
             Base of the logarithm. Natural logarithm is used by default.
@@ -393,6 +393,7 @@ def log1p(
     in-place and returns None.
 
     """
+    adata = data
     if copy:
         if not inplace:
             raise ValueError("`copy=True` cannot be used with `inplace=False`.")

src/rapids_singlecell/preprocessing/_pca.py

@@ -69,7 +69,7 @@ def _resolve_mask_var(
 
 
 def pca(
-    adata: AnnData,
+    data: AnnData,
     n_comps: int | None = None,
     *,
     layer: str = None,
@@ -112,7 +112,7 @@ def pca(
 
     Parameters
     ----------
-    adata
+    data
         AnnData object
 
     n_comps
@@ -210,6 +210,7 @@ def pca(
                 Explained variance, equivalent to the eigenvalues of the \
                 covariance matrix.
     """
+    adata = data
     if use_highly_variable is True and "highly_variable" not in adata.var.keys():
         raise ValueError(
             "Did not find adata.var['highly_variable']. "

src/rapids_singlecell/preprocessing/_scale.py

@@ -24,7 +24,7 @@
 
 
 def scale(
-    adata: AnnData,
+    data: AnnData,
     *,
     zero_center: bool = True,
     max_value: float | None = None,
@@ -39,7 +39,7 @@ def scale(
 
     Parameters
     ----------
-        adata
+        data
             AnnData object
 
         zero_center
@@ -74,6 +74,7 @@ def scale(
     depending on `inplace`.
 
     """
+    adata = data
     if copy:
         if not inplace:
             raise ValueError("`copy=True` cannot be used with `inplace=False`.")

src/rapids_singlecell/squidpy_gpu/_autocorr.py

@@ -12,6 +12,7 @@
 from scipy import sparse
 from statsmodels.stats.multitest import multipletests
 
+from rapids_singlecell._compat import SpatialData
 from rapids_singlecell.preprocessing._utils import _sparse_to_dense
 
 from ._gearysc import _gearys_C_cupy
@@ -49,7 +50,7 @@ def _to_cupy(vals, *, use_sparse: bool, dtype):
 
 
 def spatial_autocorr(
-    adata: AnnData,
+    adata: AnnData | SpatialData,
     *,
     connectivity_key: str = "spatial_connectivities",
     genes: str | Sequence[str] | None = None,
@@ -118,6 +119,8 @@ def spatial_autocorr(
             DataFrame containing the autocorrelation scores, p-values, and corrected p-values for each gene. \
             If `copy` is False, the results are stored in `adata.uns` and None is returned.
     """
+    if SpatialData is not None and isinstance(adata, SpatialData):
+        adata = adata.table
     if genes is None:
         if "highly_variable" in adata.var:
             genes = adata[:, adata.var["highly_variable"]].var_names.values

src/rapids_singlecell/squidpy_gpu/_co_oc.py

@@ -6,6 +6,7 @@
 import numpy as np
 from cuml.metrics import pairwise_distances
 
+from rapids_singlecell._compat import SpatialData
 from rapids_singlecell._cuda import _cooc_cuda as _co
 from rapids_singlecell._utils import (
     _calculate_blocks_per_pair,
@@ -21,7 +22,7 @@
 
 
 def co_occurrence(
-    adata: AnnData,
+    adata: AnnData | SpatialData,
     cluster_key: str,
     *,
     spatial_key: str = "spatial",
@@ -65,6 +66,8 @@ def co_occurrence(
           computed at ``interval``.
     """
 
+    if SpatialData is not None and isinstance(adata, SpatialData):
+        adata = adata.table
     _assert_categorical_obs(adata, key=cluster_key)
     _assert_spatial_basis(adata, key=spatial_key)
     spatial = cp.array(adata.obsm[spatial_key]).astype(np.float32)