datasets

src/beignet/datasets/__init__.py

@@ -11,7 +11,9 @@
 from ._atom3d_res_dataset import ATOM3DRESDataset
 from ._atom3d_rsr_dataset import ATOM3DRSRDataset
 from ._atom3d_smp_dataset import ATOM3DSMPDataset
-from ._bindingdb_dataset import BindingDBDataset
+from ._bindingdb_ic50_dataset import BindingDBIC50Dataset
+from ._bindingdb_kd_dataset import BindingDBKdDataset
+from ._bindingdb_ki_dataset import BindingDBKiDataset
 from ._broccatelli_p_glycoprotein_inhibition_dataset import (
     BroccatelliPGlycoproteinInhibitionDataset,
 )
@@ -98,26 +100,28 @@
 from ._zinc_dataset import ZINCDataset
 
 __all__ = [
-    "AqSolDBSolubilityDataset",
-    "AstraZenecaClearanceDataset",
-    "AstraZenecaLipophilicityDataset",
-    "AstraZenecaPlasmaProteinBindingRateDataset",
     "ATOM3DDataset",
     "ATOM3DMSPDataset",
     "ATOM3DPPIDataset",
     "ATOM3DPSRDataset",
     "ATOM3DRESDataset",
     "ATOM3DRSRDataset",
     "ATOM3DSMPDataset",
-    "BindingDBDataset",
+    "AqSolDBSolubilityDataset",
+    "AstraZenecaClearanceDataset",
+    "AstraZenecaLipophilicityDataset",
+    "AstraZenecaPlasmaProteinBindingRateDataset",
+    "BindingDBIC50Dataset",
+    "BindingDBKdDataset",
+    "BindingDBKiDataset",
     "BroccatelliPGlycoproteinInhibitionDataset",
     "CarbonMangelsCytochromeP4502C9SubstrateDataset",
     "CarbonMangelsCytochromeP4502D6SubstrateDataset",
     "CarbonMangelsCytochromeP4503A4SubstrateDataset",
     "ChEMBLDataset",
     "ClinToxDataset",
-    "DataFrameDataset",
     "DAVISDataset",
+    "DataFrameDataset",
     "DisGeNETDataset",
     "DrugCombDataset",
     "FASTADataset",
@@ -130,46 +134,46 @@
     "KIBADataset",
     "LMDBDataset",
     "LombardoVolumeOfDistributionAtSteadyStateDataset",
+    "MOSESDataset",
     "MaBioavailabilityDataset",
     "MartinsBloodBrainBarrierDataset",
-    "MOSESDataset",
     "NCATSPAMPAPermeabilityDataset",
     "ObachHalfLifeDataset",
-    "ParquetDataset",
     "PDB70Dataset",
     "PDBTrajectoryDataset",
     "PDBbindDataset",
+    "ParquetDataset",
     "QM7Dataset",
     "QM7bDataset",
     "QM8Dataset",
     "QM9Dataset",
+    "REALDatabaseDataset",
     "RandomEulerAngleDataset",
     "RandomQuaternionDataset",
     "RandomRotationDataset",
     "RandomRotationMatrixDataset",
     "RandomRotationVectorDataset",
-    "REALDatabaseDataset",
     "SAbDabDataset",
+    "SKEMPIDataset",
     "SequenceDataset",
     "SizedSequenceDataset",
-    "SKEMPIDataset",
     "SwissProtDataset",
     "TDCDataset",
     "Tox21Dataset",
     "ToxCastDataset",
-    "TrajectoryDataset",
     "TrEMBLDataset",
+    "TrajectoryDataset",
+    "USPTOReactionProductDataset",
     "UniProtDataset",
     "UniRef100Dataset",
     "UniRef50Dataset",
     "UniRef90Dataset",
-    "USPTOReactionProductDataset",
     "VeithCytochromeP4501A2InhibitionDataset",
     "VeithCytochromeP4502C19InhibitionDataset",
     "VeithCytochromeP4502C9InhibitionDataset",
     "VeithCytochromeP4502D6InhibitionDataset",
     "VeithCytochromeP4503A4InhibitionDataset",
     "WangEffectivePermeabilityDataset",
-    "ZhuAcuteToxicityLD50Dataset",
     "ZINCDataset",
+    "ZhuAcuteToxicityLD50Dataset",
 ]

src/beignet/datasets/_bindingdb_ic50_dataset.py

@@ -0,0 +1,45 @@
+from pathlib import Path
+from typing import Callable
+
+from beignet.transforms import Transform
+
+from ._tdc_dataset import TDCDataset
+
+
+class BindingDBIC50Dataset(TDCDataset):
+    def __init__(
+        self,
+        root: str | Path,
+        *,
+        download: bool = False,
+        transform: Callable | Transform | None = None,
+        target_transform: Callable | Transform | None = None,
+    ):
+        r"""
+
+        Parameters
+        ----------
+        root : str | Path
+            Root directory of dataset.
+
+        download: bool
+            If `True`, downloads the dataset to the root directory. If dataset
+            already exists, it is not redownloaded. Default, `False`.
+
+        transform : Callable | Transform | None
+            Transforms the input.
+
+        target_transform : Callable | Transform | None
+            Transforms the target.
+        """
+        super().__init__(
+            root=root,
+            download=download,
+            identifier=4291560,
+            suffix="csv",
+            checksum="md5:a6ca198002c335aa9a30248cf3795413",
+            x_keys=["X1", "X2"],
+            y_keys=["Y"],
+            transform=transform,
+            target_transform=target_transform,
+        )

src/beignet/datasets/_bindingdb_dataset.py → src/beignet/datasets/_bindingdb_kd_dataset.py

@@ -6,7 +6,7 @@
 from ._tdc_dataset import TDCDataset
 
 
-class BindingDBDataset(TDCDataset):
+class BindingDBKdDataset(TDCDataset):
     def __init__(
         self,
         root: str | Path,
@@ -35,11 +35,11 @@ def __init__(
         super().__init__(
             root=root,
             download=download,
-            identifier=0,
-            suffix="",
-            checksum="",
-            x_keys=[""],
-            y_keys=[""],
+            identifier=4291555,
+            suffix="csv",
+            checksum="md5:728c9d39485cf9667c1567cc710fb6b6",
+            x_keys=["X1", "X2"],
+            y_keys=["Y"],
             transform=transform,
             target_transform=target_transform,
         )

src/beignet/datasets/_bindingdb_ki_dataset.py

@@ -0,0 +1,45 @@
+from pathlib import Path
+from typing import Callable
+
+from beignet.transforms import Transform
+
+from ._tdc_dataset import TDCDataset
+
+
+class BindingDBKiDataset(TDCDataset):
+    def __init__(
+        self,
+        root: str | Path,
+        *,
+        download: bool = False,
+        transform: Callable | Transform | None = None,
+        target_transform: Callable | Transform | None = None,
+    ):
+        r"""
+
+        Parameters
+        ----------
+        root : str | Path
+            Root directory of dataset.
+
+        download: bool
+            If `True`, downloads the dataset to the root directory. If dataset
+            already exists, it is not redownloaded. Default, `False`.
+
+        transform : Callable | Transform | None
+            Transforms the input.
+
+        target_transform : Callable | Transform | None
+            Transforms the target.
+        """
+        super().__init__(
+            root=root,
+            download=download,
+            identifier=4291556,
+            suffix="csv",
+            checksum="md5:187d6c8926c608e24f4469373811806d",
+            x_keys=["X1", "X2"],
+            y_keys=["Y"],
+            transform=transform,
+            target_transform=target_transform,
+        )

src/beignet/datasets/_davis_dataset.py

@@ -35,11 +35,11 @@ def __init__(
         super().__init__(
             root=root,
             download=download,
-            identifier=0,
-            suffix="",
+            identifier=5219748,
+            suffix="md5:6c7949b81aea69e9d816db88602d771e",
             checksum="",
-            x_keys=[""],
-            y_keys=[""],
+            x_keys=["X1", "X2"],
+            y_keys=["Y"],
             transform=transform,
             target_transform=target_transform,
         )

src/beignet/datasets/_kiba_dataset.py

@@ -35,11 +35,11 @@ def __init__(
         super().__init__(
             root=root,
             download=download,
-            identifier=0,
-            suffix="",
-            checksum="",
-            x_keys=[""],
-            y_keys=[""],
+            identifier=5255037,
+            suffix="tsv",
+            checksum="md5:2af3d738f320700bfc98b781fe354b64",
+            x_keys=["X1", "X2"],
+            y_keys=["Y"],
             transform=transform,
             target_transform=target_transform,
         )