pertpy 0.11.5__tar.gz → 1.0.0__tar.gz

{pertpy-0.11.5 → pertpy-1.0.0}/.github/release-drafter.yml

@@ -1,5 +1,5 @@
-name-template: "0.11.5 🌈"
-tag-template: 0.11.5
+name-template: "1.0.0 🌈"
+tag-template: 1.0.0
 exclude-labels:
     - "skip-changelog"
 

{pertpy-0.11.5 → pertpy-1.0.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pertpy
-Version: 0.11.5
+Version: 1.0.0
 Summary: Perturbation Analysis in the scverse ecosystem.
 Project-URL: Documentation, https://pertpy.readthedocs.io
 Project-URL: Source, https://github.com/scverse/pertpy
@@ -155,7 +155,13 @@ pip install 'pertpy[tcoda]'
 
 ### milo
 
-milo further requires edger, statmod, and rpy2 to be installed:
+milo requires either the "de" extra for the "pydeseq2" solver:
+
+```console
+pip install 'pertpy[de]'
+```
+
+or, edger, statmod, and rpy2 for the "edger" solver:
 
 ```R
 BiocManager::install("edgeR")

{pertpy-0.11.5 → pertpy-1.0.0}/README.md

@@ -53,7 +53,13 @@ pip install 'pertpy[tcoda]'
 
 ### milo
 
-milo further requires edger, statmod, and rpy2 to be installed:
+milo requires either the "de" extra for the "pydeseq2" solver:
+
+```console
+pip install 'pertpy[de]'
+```
+
+or, edger, statmod, and rpy2 for the "edger" solver:
 
 ```R
 BiocManager::install("edgeR")

{pertpy-0.11.5 → pertpy-1.0.0}/docs/changelog.md

@@ -5,7 +5,28 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
-## v0.11.3
+## v1.0.0
+
+### 🚀 Features
+
+* Simplify guide assignment tutorial ([#796](https://github.com/scverse/pertpy/pull/796)) @Zethson
+* Add milo pydeseq2 solver ([#795](https://github.com/scverse/pertpy/pull/795)) @Zethson
+
+## v0.11.5
+
+### 🚀 Features
+
+* Move to trusted publishing ([#790](https://github.com/scverse/pertpy/pull/790)) @Zethson
+* Add conda-forge to installation instructions ([#789](https://github.com/scverse/pertpy/pull/789)) @Zethson
+* Fix seaborn 0.14 compat & testing dependencies & speed up tests ([#787](https://github.com/scverse/pertpy/pull/787)) @Zethson
+* Support > 2 colors in `pt.tl.Sccoda.plot_boxplots` ([#784](https://github.com/scverse/pertpy/pull/784)) @mschilli87
+
+### 🐛 Bug Fixes
+
+* Fix LIL support for AnnData 0.12 ([#792](https://github.com/scverse/pertpy/pull/792)) @Zethson
+* Fix compatibility with rpy 3.6.0 in milo ([#791](https://github.com/scverse/pertpy/pull/791)) @seohyonkim
+
+## v0.11.4
 
 ### 🧰 Maintenance
 
@@ -13,7 +34,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## v0.11.3
 
-## 🚀 Features
+### 🚀 Features
 
 * add about page ([#770](https://github.com/scverse/pertpy/pull/770)) @Zethson
 * Simplify Metadata errors ([#765](https://github.com/scverse/pertpy/pull/765)) @Zethson

{pertpy-0.11.5 → pertpy-1.0.0}/docs/index.md

@@ -48,7 +48,7 @@ It provides tools for harmonizing perturbation datasets, automating metadata ann
 ```{toctree}
 :caption: 'General'
 :hidden: true
-:maxdepth: 1
+:maxdepth: 2
 
 installation
 api
@@ -60,7 +60,7 @@ references
 ```{toctree}
 :caption: 'Gallery'
 :hidden: true
-:maxdepth: 1
+:maxdepth: 3
 
 tutorials
 usecases
@@ -69,7 +69,7 @@ usecases
 ```{toctree}
 :caption: 'About'
 :hidden: true
-:maxdepth: 1
+:maxdepth: 2
 
 about/background
 about/cite

{pertpy-0.11.5 → pertpy-1.0.0}/docs/installation.md

@@ -47,7 +47,13 @@ pip install rpy2
 
 #### milo
 
-milo further requires edger, statmod, and rpy2 to be installed:
+milo requires either the "de" extra for the "pydeseq2" solver:
+
+```console
+pip install 'pertpy[de]'
+```
+
+or, edger, statmod, and rpy2 for the "edger" solver:
 
 ```R
 BiocManager::install("edgeR")

{pertpy-0.11.5 → pertpy-1.0.0}/pertpy/__init__.py

@@ -2,10 +2,11 @@
 
 __author__ = "Lukas Heumos"
 __email__ = "lukas.heumos@posteo.net"
-__version__ = "0.11.5"
+__version__ = "1.0.0"
 
 import warnings
 
+from anndata._core.aligned_df import ImplicitModificationWarning
 from matplotlib import MatplotlibDeprecationWarning
 from numba import NumbaDeprecationWarning
 
@@ -13,6 +14,8 @@ warnings.filterwarnings("ignore", category=NumbaDeprecationWarning)
 warnings.filterwarnings("ignore", category=MatplotlibDeprecationWarning)
 warnings.filterwarnings("ignore", category=SyntaxWarning)
 warnings.filterwarnings("ignore", category=UserWarning, module="scvi._settings")
+warnings.filterwarnings("ignore", message="Environment variable.*redefined by R")
+warnings.filterwarnings("ignore", message="Transforming to str index.", category=ImplicitModificationWarning)
 
 import mudata
 

{pertpy-0.11.5 → pertpy-1.0.0}/pertpy/tools/_milo.py

@@ -2,6 +2,7 @@ from __future__ import annotations
 
 import random
 import re
+from importlib.util import find_spec
 from typing import TYPE_CHECKING, Literal
 
 import matplotlib.pyplot as plt
@@ -29,18 +30,6 @@ from sklearn.metrics.pairwise import euclidean_distances
 class Milo:
     """Python implementation of Milo."""
 
-    def __init__(self):
-        try:
-            from rpy2.robjects import conversion, numpy2ri, pandas2ri
-            from rpy2.robjects.packages import STAP, PackageNotInstalledError, importr
-        except ModuleNotFoundError:
-            raise ImportError("milo requires rpy2 to be installed.") from None
-
-        try:
-            importr("edgeR")
-        except ImportError as e:
-            raise ImportError("milo requires a valid R installation with edger installed:\n") from e
-
     def load(
         self,
         input: AnnData,
@@ -266,7 +255,7 @@ class Milo:
         subset_samples: list[str] | None = None,
         add_intercept: bool = True,
         feature_key: str | None = "rna",
-        solver: Literal["edger", "batchglm"] = "edger",
+        solver: Literal["edger", "pydeseq2"] = "edger",
     ):
         """Performs differential abundance testing on neighbourhoods using QLF test implementation as implemented in edgeR.
 
@@ -279,7 +268,9 @@ class Milo:
             subset_samples: subset of samples (obs in `milo_mdata['milo']`) to use for the test.
             add_intercept: whether to include an intercept in the model. If False, this is equivalent to adding + 0 in the design formula. When model_contrasts is specified, this is set to False by default.
             feature_key: If input data is MuData, specify key to cell-level AnnData object.
-            solver: The solver to fit the model to. One of "edger" (requires R, rpy2 and edgeR to be installed) or "batchglm"
+            solver: The solver to fit the model to.
+                The "edger" solver requires R, rpy2 and edgeR to be installed and is the closest to the R implementation.
+                The "pydeseq2" requires pydeseq2 to be installed. It is still very comparable to the "edger" solver but might be a bit slower.
 
         Returns:
             None, modifies `milo_mdata['milo']` in place, adding the results of the DA test to `.var`:
@@ -298,7 +289,6 @@ class Milo:
             >>> milo.make_nhoods(mdata["rna"])
             >>> mdata = milo.count_nhoods(mdata, sample_col="orig.ident")
             >>> milo.da_nhoods(mdata, design="~label")
-
         """
         try:
             sample_adata = mdata["milo"]
@@ -425,13 +415,65 @@ class Milo:
                 res = pd.DataFrame(res)
             # The columns of res looks like e.g. table.A, table.B, so remove the prefix
             res.columns = [col.replace("table.", "") for col in res.columns]
-        # Save outputs
+        elif solver == "pydeseq2":
+            if find_spec("pydeseq2") is None:
+                raise ImportError("pydeseq2 is required but not installed. Install with: pip install pydeseq2")
+
+            from pydeseq2.dds import DeseqDataSet
+            from pydeseq2.ds import DeseqStats
+
+            counts_filtered = count_mat[np.ix_(keep_nhoods, keep_smp)]
+            design_df_filtered = design_df.iloc[keep_smp].copy()
+
+            design_df_filtered = design_df_filtered.astype(
+                dict.fromkeys(design_df_filtered.select_dtypes(exclude=["number"]).columns, "category")
+            )
+
+            design_clean = design if design.startswith("~") else f"~{design}"
+
+            dds = DeseqDataSet(
+                counts=pd.DataFrame(counts_filtered.T, index=design_df_filtered.index),
+                metadata=design_df_filtered,
+                design=design_clean,
+                refit_cooks=True,
+            )
+
+            dds.deseq2()
+
+            if model_contrasts is not None and "-" in model_contrasts:
+                if "(" in model_contrasts or "+" in model_contrasts.split("-")[1]:
+                    raise ValueError(
+                        f"Complex contrasts like '{model_contrasts}' are not supported by pydeseq2. "
+                        "Use simple pairwise contrasts (e.g., 'GroupA-GroupB') or switch to solver='edger'."
+                    )
+
+                parts = model_contrasts.split("-")
+                factor_name = design_clean.replace("~", "").split("+")[-1].strip()
+                group1 = parts[0].replace(factor_name, "").strip()
+                group2 = parts[1].replace(factor_name, "").strip()
+                stat_res = DeseqStats(dds, contrast=[factor_name, group1, group2])
+            else:
+                factor_name = design_clean.replace("~", "").split("+")[-1].strip()
+                if not isinstance(design_df_filtered[factor_name], pd.CategoricalDtype):
+                    design_df_filtered[factor_name] = design_df_filtered[factor_name].astype("category")
+                categories = design_df_filtered[factor_name].cat.categories
+                stat_res = DeseqStats(dds, contrast=[factor_name, categories[-1], categories[0]])
+
+            stat_res.summary()
+            res = stat_res.results_df
+
+            res = res.rename(
+                columns={"baseMean": "logCPM", "log2FoldChange": "logFC", "pvalue": "PValue", "padj": "FDR"}
+            )
+
+            res = res[["logCPM", "logFC", "PValue", "FDR"]]
+
         res.index = sample_adata.var_names[keep_nhoods]  # type: ignore
         if any(col in sample_adata.var.columns for col in res.columns):
             sample_adata.var = sample_adata.var.drop(res.columns, axis=1)
         sample_adata.var = pd.concat([sample_adata.var, res], axis=1)
-        # Run Graph spatial FDR correction
-        self._graph_spatial_fdr(sample_adata, neighbors_key=adata.uns["nhood_neighbors_key"])
+
+        self._graph_spatial_fdr(sample_adata)
 
     def annotate_nhoods(
         self,
@@ -674,6 +716,17 @@ class Milo:
         self,
     ):
         """Set up rpy2 to run edgeR."""
+        try:
+            from rpy2.robjects import conversion, numpy2ri, pandas2ri
+            from rpy2.robjects.packages import STAP, PackageNotInstalledError, importr
+        except ModuleNotFoundError:
+            raise ImportError("milo requires rpy2 to be installed.") from None
+
+        try:
+            importr("edgeR")
+        except ImportError as e:
+            raise ImportError("milo requires a valid R installation with edger installed.") from e
+
         from rpy2.robjects.packages import importr
 
         edgeR = self._try_import_bioc_library("edgeR")
@@ -685,26 +738,27 @@ class Milo:
 
     def _try_import_bioc_library(
         self,
-        name: str,
+        r_package: str,
     ):
         """Import R packages.
 
         Args:
-            name (str): R packages name
+            r_package: R packages name
         """
         from rpy2.robjects.packages import PackageNotInstalledError, importr
 
         try:
-            _r_lib = importr(name)
+            _r_lib = importr(r_package)
             return _r_lib
         except PackageNotInstalledError:
-            logger.error(f"Install Bioconductor library `{name!r}` first as `BiocManager::install({name!r}).`")
+            logger.error(
+                f"Install Bioconductor library `{r_package!r}` first as `BiocManager::install({r_package!r}).`"
+            )
             raise
 
     def _graph_spatial_fdr(
         self,
         sample_adata: AnnData,
-        neighbors_key: str | None = None,
     ):
         """FDR correction weighted on inverse of connectivity of neighbourhoods.
 
@@ -712,7 +766,6 @@ class Milo:
 
         Args:
             sample_adata: Sample-level AnnData.
-            neighbors_key: The key in `adata.obsp` to use as KNN graph.
         """
         # use 1/connectivity as the weighting for the weighted BH adjustment from Cydar
         w = 1 / sample_adata.var["kth_distance"]

{pertpy-0.11.5 → pertpy-1.0.0}/pertpy/tools/_perturbation_space/_perturbation_space.py

@@ -80,7 +80,7 @@ class PerturbationSpace:
         group_masks = (
             [(adata.obs[group_col] == sample) for sample in adata.obs[group_col].unique()]
             if group_col
-            else [[True] * adata.n_obs]
+            else [np.array([True] * adata.n_obs)]
         )
 
         if layer_key:

{pertpy-0.11.5 → pertpy-1.0.0}/pyproject.toml

@@ -4,7 +4,7 @@ requires = ["hatchling"]
 
 [project]
 name = "pertpy"
-version = "0.11.5"
+version = "1.0.0"
 description = "Perturbation Analysis in the scverse ecosystem."
 readme = "README.md"
 requires-python = ">=3.11,<3.14"
@@ -61,7 +61,7 @@ dependencies = [
     "pubchempy",
     "pyarrow",
     "blitzgsea",
-    "scikit-learn>=1.4",  # https://github.com/scverse/pertpy/issues/700
+    "scikit-learn>=1.4",
     "fast-array-utils",
     "lamin_utils",
     "arviz",

{pertpy-0.11.5 → pertpy-1.0.0}/tests/tools/test_milo.py

@@ -1,3 +1,5 @@
+from importlib.util import find_spec
+
 import numpy as np
 import pandas as pd
 import pertpy as pt
@@ -5,14 +7,23 @@ import pytest
 import scanpy as sc
 from mudata import MuData
 
-try:
-    from rpy2.robjects.packages import importr
 
-    r_dependency = importr("edgeR")
-except Exception:  # noqa: BLE001
-    r_dependency = None
+@pytest.fixture(params=["edger", "pydeseq2"])
+def solver(request):
+    solver_name = request.param
+
+    if solver_name == "edger":
+        try:
+            from rpy2.robjects.packages import importr
+
+            importr("edgeR")
+        except Exception:  # noqa: BLE001
+            pytest.skip("Required R package 'edgeR' not available")
 
-pytestmark = pytest.mark.skipif(r_dependency is None, reason="Required R package 'edgeR' not available")
+    elif solver_name == "pydeseq2" and find_spec("pydeseq2") is None:
+        pytest.skip("pydeseq2 not available")
+
+    return solver_name
 
 
 @pytest.fixture
@@ -136,6 +147,23 @@ def da_nhoods_mdata(adata, milo):
     return milo_mdata
 
 
+def test_da_nhoods_pvalues_both_solvers(da_nhoods_mdata, milo, solver):
+    mdata = da_nhoods_mdata.copy()
+    milo.da_nhoods(mdata, design="~condition", solver=solver)
+    sample_adata = mdata["milo"].copy()
+    min_p, max_p = sample_adata.var["PValue"].min(), sample_adata.var["PValue"].max()
+    assert (min_p >= 0) & (max_p <= 1), "P-values are not between 0 and 1"
+
+
+def test_da_nhoods_fdr_both_solvers(da_nhoods_mdata, milo, solver):
+    mdata = da_nhoods_mdata.copy()
+    milo.da_nhoods(mdata, design="~condition", solver=solver)
+    sample_adata = mdata["milo"].copy()
+    assert np.all(np.round(sample_adata.var["PValue"], 10) <= np.round(sample_adata.var["SpatialFDR"], 10)), (
+        "FDR is higher than uncorrected P-values"
+    )
+
+
 def test_da_nhoods_missing_samples(adata, milo):
     with pytest.raises(KeyError):
         milo.da_nhoods(adata, design="~condition")

{pertpy-0.11.5 → pertpy-1.0.0}/tests/tools/test_scgen.py

@@ -1,3 +1,5 @@
+import warnings
+
 import anndata as ad
 import pertpy as pt
 import scanpy as sc
@@ -5,8 +7,10 @@ from scvi.data import synthetic_iid
 
 
 def test_scgen():
-    adata = synthetic_iid()
-    adata.obs_names_make_unique()
+    with warnings.catch_warnings():
+        warnings.filterwarnings("ignore", message="Observation names are not unique")
+        adata = synthetic_iid()
+        adata.obs_names_make_unique()
     pt.tl.Scgen.setup_anndata(
         adata,
         batch_key="batch",