Merge branch 'main' into fix-hash-for-chemprop

Author: c-w-feldmann

.pre-commit-config.yaml

@@ -24,7 +24,7 @@ repos:
         stages: [commit-msg]
 
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.15.7
+    rev: v0.15.14
     hooks:
       - id: ruff-format
         types_or: [python, pyi, jupyter]

molpipeline/estimators/chemprop/component_wrapper.py

@@ -220,12 +220,9 @@ def set_params(self, **params: Any) -> Self:
 
 
 # pylint: disable=too-many-ancestors, too-many-instance-attributes
-class PredictorWrapper(_Predictor, BaseEstimator, abc.ABC):  # type: ignore
+class PredictorWrapper(_Predictor, BaseEstimator, abc.ABC):
     """Abstract wrapper for the Predictor class."""
 
-    _T_default_criterion: ChempropMetric
-    _T_default_metric: ChempropMetric
-
     def __init__(  # pylint: disable=too-many-positional-arguments  # noqa: PLR0917
         self,
         n_tasks: int = 1,
@@ -384,45 +381,45 @@ def set_params(self, **params: Any) -> Self:
         return self
 
 
-class RegressionFFN(PredictorWrapper, _RegressionFFN):  # type: ignore
+class RegressionFFN(PredictorWrapper, _RegressionFFN):
     """A wrapper for the RegressionFFN class."""
 
     n_targets: int = 1
     _T_default_criterion = MSE
     _T_default_metric = MSE
 
 
-class MveFFN(PredictorWrapper, _MveFFN):  # type: ignore
+class MveFFN(PredictorWrapper, _MveFFN):
     """A wrapper for the MveFFN class."""
 
     n_targets: int = 2
     _T_default_criterion = MVELoss
 
 
-class EvidentialFFN(PredictorWrapper, _EvidentialFFN):  # type: ignore
+class EvidentialFFN(PredictorWrapper, _EvidentialFFN):
     """A wrapper for the EvidentialFFN class."""
 
     n_targets: int = 4
     _T_default_criterion = EvidentialLoss
 
 
-class BinaryClassificationFFN(PredictorWrapper, _BinaryClassificationFFN):  # type: ignore
+class BinaryClassificationFFN(PredictorWrapper, _BinaryClassificationFFN):
     """A wrapper for the BinaryClassificationFFN class."""
 
     n_targets: int = 1
     _T_default_criterion = BCELoss
     _T_default_metric = BinaryAUROC
 
 
-class BinaryDirichletFFN(PredictorWrapper, _BinaryDirichletFFN):  # type: ignore
+class BinaryDirichletFFN(PredictorWrapper, _BinaryDirichletFFN):
     """A wrapper for the BinaryDirichletFFN class."""
 
     n_targets: int = 2
     _T_default_criterion = DirichletLoss
     _T_default_metric = BinaryAUROC
 
 
-class MulticlassClassificationFFN(PredictorWrapper, _MulticlassClassificationFFN):  # type: ignore
+class MulticlassClassificationFFN(PredictorWrapper, _MulticlassClassificationFFN):
     """A wrapper for the MulticlassClassificationFFN class."""
 
     n_targets: int = 1
@@ -508,15 +505,15 @@ def __setstate__(self, state: dict[str, Any]) -> None:
         super().__setstate__(state)
 
 
-class MulticlassDirichletFFN(PredictorWrapper, _MulticlassDirichletFFN):  # type: ignore
+class MulticlassDirichletFFN(PredictorWrapper, _MulticlassDirichletFFN):
     """A wrapper for the MulticlassDirichletFFN class."""
 
     n_targets: int = 1
     _T_default_criterion = DirichletLoss
     _T_default_metric = MulticlassMCCMetric
 
 
-class SpectralFFN(PredictorWrapper, _SpectralFFN):  # type: ignore
+class SpectralFFN(PredictorWrapper, _SpectralFFN):
     """A wrapper for the SpectralFFN class."""
 
     n_targets: int = 1

molpipeline/estimators/chemprop/lightning_wrapper.py

@@ -146,7 +146,7 @@ def get_params_trainer(trainer: pl.Trainer) -> dict[str, Any]:
         "num_nodes": trainer.num_nodes,
         "precision": trainer.precision,
         "logger": trainer.logger,
-        # "callbacks": trainer.callbacks,  # type: ignore[attr-defined]
+        # "callbacks": trainer.callbacks,  # type: ignore[attr-defined]  # noqa: ERA001
         "fast_dev_run": trainer.fast_dev_run,  # type: ignore[attr-defined]
         "max_epochs": trainer.max_epochs,
         "min_epochs": trainer.min_epochs,
@@ -172,11 +172,11 @@ def get_params_trainer(trainer: pl.Trainer) -> dict[str, Any]:
         "benchmark": torch.backends.cudnn.benchmark,
         "inference_mode": trainer.predict_loop.inference_mode,
         "use_distributed_sampler": trainer._accelerator_connector.use_distributed_sampler,  # noqa: SLF001,E501
-        # "profiler": trainer.profiler,  # type: ignore[attr-defined]
+        # "profiler": trainer.profiler,  # type: ignore[attr-defined]  # noqa: ERA001
         "detect_anomaly": trainer._detect_anomaly,  # noqa: SLF001
         "barebones": trainer.barebones,
-        # "plugins": trainer.plugins,  # can not be exctracted
-        # "sync_batchnorm": trainer._accelerator_connector.sync_batchnorm,  # plugin related  # noqa: E501
+        # "plugins": trainer.plugins,  # can not be exctracted  # noqa: ERA001
+        # "sync_batchnorm": trainer._accelerator_connector.sync_batchnorm,  # plugin related  # noqa: E501, ERA001
         "reload_dataloaders_every_n_epochs": trainer.reload_dataloaders_every_n_epochs,  # type: ignore[attr-defined]
         "default_root_dir": get_trainer_path(trainer),
     }

molpipeline/estimators/leader_picker_clustering.py

@@ -120,7 +120,7 @@ def _fit(self, X: list[ExplicitBitVect]) -> Self:  # noqa: N803
             poolSize=len(X),
             threshold=self.distance_threshold,
             numThreads=1,  # according to rdkit docu this parameter is not used
-            # seed=self.random_state if self.random_state is not None else -1,
+            # seed=self.random_state if self.random_state is not None else -1,  # noqa: ERA001
         )
 
         # Assign points to clusters based on centroid

molpipeline/pipeline/_skl_pipeline.py

@@ -1021,7 +1021,7 @@ def __sklearn_tags__(self) -> Tags:
             # tuples and `fit` is not called yet to validate the steps.
             pass
 
-        try:
+        try:  # noqa: PLW0717
             # Only the _final_estimator is changed from the original implementation is
             # changed in the following 2 lines
             if (

molpipeline/utils/logging.py

@@ -69,14 +69,16 @@ def print_elapsed_time(
         yield
     else:
         start = timeit.default_timer()
-        yield
-        message_to_print = _message_with_time(
-            source,
-            message,
-            timeit.default_timer() - start,
-        )
-
-        if use_logger:
-            logger.info(message_to_print)
-        else:
-            print(message_to_print)  # noqa: T201
+        try:
+            yield
+        finally:
+            message_to_print = _message_with_time(
+                source,
+                message,
+                timeit.default_timer() - start,
+            )
+
+            if use_logger:
+                logger.info(message_to_print)
+            else:
+                print(message_to_print)  # noqa: T201

notebooks/advanced_01_hyperopt_on_bbbp.ipynb

@@ -391,8 +391,7 @@
     "plt.xlabel(\"Cluster rank by size\")\n",
     "plt.ylabel(\"Number of molecules in cluster\")\n",
     "plt.yscale(\"log\")\n",
-    "_ = plt.title(\"Sorted distribution of cluster sizes\")\n",
-    "# plt.savefig(\"cluster_sizes.pdf\", dpi=300)"
+    "_ = plt.title(\"Sorted distribution of cluster sizes\")"
    ]
   },
   {
@@ -1295,8 +1294,7 @@
     "plt.ylabel(\"ROC AUC\", fontsize=13)\n",
     "plt.xlabel(\"Estimators\", fontsize=13)\n",
     "plt.xticks(fontsize=12)\n",
-    "plt.tight_layout()\n",
-    "# plt.savefig(\"hyperopt_performance.pdf\", dpi=300)"
+    "plt.tight_layout()"
    ]
   },
   {

notebooks/advanced_03_introduction_to_explainable_ai.ipynb

@@ -1074,8 +1074,6 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "# series_names = [\"b1i5k\", \"b1i2n\", \"b1i5b\"]\n",
-    "# series_dict = {k:explanations_dict[k] for k in series_names}\n",
     "series_dict = explanations_dict\n",
     "\n",
     "weight_min = min(exp.atom_weights.min() for exp in series_dict.values())\n",

notebooks/advanced_05_conformal_prediction.ipynb

@@ -716,7 +716,7 @@
     "    confidence=confidence_level,\n",
     ")\n",
     "\n",
-    "try:\n",
+    "try:  # noqa: PLW0717\n",
     "    results_cross_clf = cross_cp_clf.evaluate(\n",
     "        x_test_clf,\n",
     "        y_test_clf,\n",
@@ -1199,7 +1199,7 @@
     "intervals_split = split_cp_reg.predict_int(x_test_reg, confidence=confidence_level)\n",
     "\n",
     "# Evaluate using built-in evaluate method and helper functions\n",
-    "try:\n",
+    "try:  # noqa: PLW0717\n",
     "    results_built_in = split_cp_reg.evaluate(\n",
     "        x_test_reg,\n",
     "        y_test_reg,\n",
@@ -1289,7 +1289,7 @@
     "intervals_cross = cross_cp_reg.predict_int(x_test_reg, confidence=confidence_level)\n",
     "\n",
     "# Evaluate using built-in evaluate method\n",
-    "try:\n",
+    "try:  # noqa: PLW0717\n",
     "    results_cross_built_in = cross_cp_reg.evaluate(\n",
     "        x_test_reg,\n",
     "        y_test_reg,\n",

pyproject.toml

@@ -91,7 +91,6 @@ ignore = [
     "ANN401",  #  Allow  typing.Any
     "D203", # incorrect-blank-line-before-class
     "D213", # blank-line-before-class-docstring
-    "ERA001",  # Found commented-out code  # TO REMOVE
     "FURB113", # Use `xxx.extend(...)` instead of repeatedly calling `xxx.append()  # TO REMOVE
     "N813",  #  Camelcase `rdFingerprintGenerator` imported as lowercase `rdkit_fp`  # TO REMOVE
     "PGH003",  # Blanket type ignore for types
@@ -143,10 +142,10 @@ reportArgumentType = "warning"
 reportCallIssue = "warning"
 reportGeneralTypeIssues = "warning"
 reportIncompatibleMethodOverride = "warning"
-reportIncompatibleVariableOverride = "warning"
+reportIncompatibleVariableOverride = "error"
 reportInvalidTypeForm = "warning"
 reportIndexIssue = "warning"
-reportMissingImports = "warning"
+reportMissingImports = "error"
 reportMissingTypeStubs = "warning"
 reportOptionalMemberAccess = "warning"
 reportOptionalSubscript = "warning"
@@ -155,7 +154,7 @@ reportPossiblyUnboundVariable = "warning"
 reportPrivateImportUsage = "warning"
 reportRedeclaration = "warning"
 reportReturnType = "warning"
-reportUnhashable = "warning"
+reportUnhashable = "error"
 venvPath = "."
 venv = ".venv"
 pythonVersion = "3.11"
@@ -174,6 +173,7 @@ exclude = ["tests", "docs"]
 dev = [
     "bandit>=1.8.3",
     "black[jupyter]>=25.1.0",
+    "chemprop>=2.2.3",
     "coverage>=7.8.0",
     "docsig>=0.79.0",
     "flake8>=7.2.0",