Merge branch 'main' into fix/sparsify-row-materializer

Author: selmanozleyen

docs/conf.py

@@ -172,6 +172,9 @@
     r"https://doi.org/10.3390/a13090212",
     r"https://www.mdpi.com/1999-4893/13/9/212",
     r"https://pubmed\.ncbi\.nlm\.nih\.gov/.*",
+    # 404 Client Error (page moved upstream; linked from a tutorial in the `docs/notebooks` submodule)
+    r"https://muon\.readthedocs\.io/en/latest/api/generated/muon\.atac\.tl\.lsi\.html",
+    r"https://openproblems\.bio/competitions/neurips_2021_docs/data/dataset\.html",
 ]
 
 exclude_patterns = ["_build", "**.ipynb_checkpoints", "notebooks/README.rst", "notebooks/CONTRIBUTING.rst"]

pyproject.toml

@@ -237,8 +237,7 @@ ignore_roles = [
 
 [tool.mypy]
 mypy_path = "$MYPY_CONFIG_FILE_DIR/src"
-python_version = "3.10"
-plugins = "numpy.typing.mypy_plugin"
+python_version = "3.12"
 
 ignore_errors = false
 

src/moscot/__init__.py

@@ -4,9 +4,9 @@
 
 try:
     md = metadata.metadata(__name__)
-    __version__ = md.get("version", "")  # type: ignore[attr-defined]
-    __author__ = md.get("Author", "")  # type: ignore[attr-defined]
-    __maintainer__ = md.get("Maintainer-email", "")  # type: ignore[attr-defined]
+    __version__ = md.get("version", "")
+    __author__ = md.get("Author", "")
+    __maintainer__ = md.get("Maintainer-email", "")
 except ImportError:
     md = None
 

src/moscot/base/cost.py

@@ -58,7 +58,7 @@ def __call__(self, *args: Any, **kwargs: Any) -> ArrayLike:
                 f"Cost matrix contains `{np.sum(np.isnan(cost))}` NaN values, "
                 f"setting them to the maximum value `{maxx}`."
             )
-            cost = np.nan_to_num(cost, nan=maxx)  # type: ignore[arg-type, type-var]
+            cost = np.nan_to_num(cost, nan=maxx)  # type: ignore[arg-type]
         if np.any(cost < 0):
             raise ValueError(f"Cost matrix contains `{np.sum(cost < 0)}` negative values.")
         return cost

src/moscot/base/problems/_mixins.py

@@ -285,7 +285,7 @@ def _annotation_mapping(
                         key_added=None,
                     )
                     v = np.array(tm_batch.argmax(0))
-                    out.extend(source_df[annotation_label][v[i]] for i in range(len(v)))
+                    out.extend(source_df[annotation_label].iloc[v[i]] for i in range(len(v)))
 
             else:
                 target_df = _get_df_cell_transition(
@@ -309,7 +309,7 @@ def _annotation_mapping(
                         key_added=None,
                     )
                     v = np.array(tm_batch.argmax(0))
-                    out.extend(target_df[annotation_label][v[i]] for i in range(len(v)))
+                    out.extend(target_df[annotation_label].iloc[v[i]] for i in range(len(v)))
             categories = pd.Categorical(out)
             return pd.DataFrame(categories, columns=[annotation_label])
         raise NotImplementedError(f"Mapping mode `{mapping_mode!r}` is not yet implemented.")

src/moscot/datasets.py

@@ -49,7 +49,7 @@ def mosta(
     force_download
         Whether to force-download the data.
     kwargs
-        Keyword arguments for :func:`scanpy.read`.
+        Keyword arguments for :func:`~anndata.io.read_h5ad`.
 
     Returns
     -------
@@ -86,7 +86,7 @@ def hspc(
     force_download
         Whether to force-download the data.
     kwargs
-        Keyword arguments for :func:`scanpy.read`.
+        Keyword arguments for :func:`~anndata.io.read_h5ad`.
 
     Returns
     -------
@@ -125,7 +125,7 @@ def drosophila(
     force_download
         Whether to force-download the data.
     kwargs
-        Keyword arguments for :func:`scanpy.read`.
+        Keyword arguments for :func:`~anndata.io.read_h5ad`.
 
     Returns
     -------
@@ -169,7 +169,7 @@ def c_elegans(
     force_download
         Whether to force-download the data.
     kwargs
-        Keyword arguments for :func:`scanpy.read`.
+        Keyword arguments for :func:`~anndata.io.read_h5ad`.
 
     Returns
     -------
@@ -207,7 +207,7 @@ def zebrafish(
     force_download
         Whether to force-download the data.
     kwargs
-        Keyword arguments for :func:`scanpy.read`.
+        Keyword arguments for :func:`~anndata.io.read_h5ad`.
 
     Returns
     -------
@@ -250,7 +250,7 @@ def bone_marrow(
     force_download
         Whether to force-download the data.
     kwargs
-        Keyword arguments for :func:`scanpy.read`.
+        Keyword arguments for :func:`~anndata.io.read_h5ad`.
 
     Returns
     -------
@@ -345,7 +345,7 @@ def tedsim(
     force_download
         Whether to force-download the data.
     kwargs
-        Keyword arguments for :func:`scanpy.read`.
+        Keyword arguments for :func:`~anndata.io.read_h5ad`.
 
     Returns
     -------
@@ -378,7 +378,7 @@ def sciplex(
     force_download
         Whether to force-download the data.
     kwargs
-        Keyword arguments for :func:`scanpy.read`.
+        Keyword arguments for :func:`~anndata.io.read_h5ad`.
 
     Returns
     -------
@@ -408,7 +408,7 @@ def sim_align(
     force_download
         Whether to force-download the data.
     kwargs
-        Keyword arguments for :func:`scanpy.read`.
+        Keyword arguments for :func:`~anndata.io.read_h5ad`.
 
     Returns
     -------

src/moscot/problems/time/_mixins.py

@@ -806,7 +806,7 @@ def compute_batch_distances(
                     **kwargs,
                 )
             )
-        return np.mean(dist)  # type: ignore[return-value]
+        return np.mean(dist)
 
     # TODO(@MUCDK) possibly offer two alternatives, once exact EMD with POT backend and once approximate,
     # faster with same solver as used for original problems

tests/problems/base/test_general_problem.py

@@ -9,6 +9,7 @@
 from ott.geometry.pointcloud import PointCloud
 from ott.solvers.linear import solve as sinkhorn
 
+import anndata as ad
 import scanpy as sc
 from anndata import AnnData
 
@@ -205,7 +206,7 @@ def test_set_graph_xy(self, adata_x: AnnData, adata_y: AnnData, ts: Tuple[Option
         new_obs_names = [name + "_src" for name in adata_x.obs_names]
         adata_x.obs_names = new_obs_names
 
-        adata_concat = adata_x.concatenate(adata_y, index_unique=None)
+        adata_concat = ad.concat([adata_x, adata_y], index_unique=None)
         sc.pp.neighbors(adata_concat, n_neighbors=15)
         graph_to_set = pd.DataFrame(
             index=adata_concat.obs_names,
@@ -323,7 +324,7 @@ def test_set_graph_xy_test_t(self, adata_x: AnnData, adata_y: AnnData, t: float)
         new_obs_names = [name + "_src" for name in adata_x.obs_names]
         adata_x.obs_names = new_obs_names
 
-        adata_concat = adata_x.concatenate(adata_y, index_unique=None)
+        adata_concat = ad.concat([adata_x, adata_y], index_unique=None)
         sc.pp.neighbors(adata_concat, n_neighbors=5)
         graph_to_set = pd.DataFrame(
             index=adata_concat.obs_names,