datasets (#43)

Genentech · Jul 19, 2024 · e102447 · e102447
1 parent f309427
commit e102447
Show file tree

Hide file tree

Showing 96 changed files with 3,766 additions and 101 deletions.
diff --git a/docs/reference/beignet.datasets.md b/docs/reference/beignet.datasets.md
@@ -0,0 +1,80 @@
+::: beignet.datasets.ATOM3DDataset
+::: beignet.datasets.ATOM3DMSPDataset
+::: beignet.datasets.ATOM3DPPIDataset
+::: beignet.datasets.ATOM3DPSRDataset
+::: beignet.datasets.ATOM3DRESDataset
+::: beignet.datasets.ATOM3DRSRDataset
+::: beignet.datasets.ATOM3DSMPDataset
+::: beignet.datasets.AqSolDBSolubilityDataset
+::: beignet.datasets.AstraZenecaClearanceDataset
+::: beignet.datasets.AstraZenecaLipophilicityDataset
+::: beignet.datasets.AstraZenecaPlasmaProteinBindingRateDataset
+::: beignet.datasets.BindingDBIC50Dataset
+::: beignet.datasets.BindingDBKdDataset
+::: beignet.datasets.BindingDBKiDataset
+::: beignet.datasets.BroccatelliPGlycoproteinInhibitionDataset
+::: beignet.datasets.CarbonMangelsCytochromeP4502C9SubstrateDataset
+::: beignet.datasets.CarbonMangelsCytochromeP4502D6SubstrateDataset
+::: beignet.datasets.CarbonMangelsCytochromeP4503A4SubstrateDataset
+::: beignet.datasets.ChEMBLDataset
+::: beignet.datasets.ClinToxDataset
+::: beignet.datasets.DAVISDataset
+::: beignet.datasets.DataFrameDataset
+::: beignet.datasets.DisGeNETDataset
+::: beignet.datasets.DrugCombDataset
+::: beignet.datasets.FASTADataset
+::: beignet.datasets.FreeSolvDataset
+::: beignet.datasets.GDSC1Dataset
+::: beignet.datasets.GDSC2Dataset
+::: beignet.datasets.HDF5TrajectoryDataset
+::: beignet.datasets.HouHumanIntestinalAbsorptionDataset
+::: beignet.datasets.HuRIDataset
+::: beignet.datasets.JespersenIEDBDataset
+::: beignet.datasets.JespersenPDBDataset
+::: beignet.datasets.KIBADataset
+::: beignet.datasets.LMDBDataset
+::: beignet.datasets.LombardoVolumeOfDistributionAtSteadyStateDataset
+::: beignet.datasets.MOSESDataset
+::: beignet.datasets.MaBioavailabilityDataset
+::: beignet.datasets.MartinsBloodBrainBarrierDataset
+::: beignet.datasets.NCATSPAMPAPermeabilityDataset
+::: beignet.datasets.ObachHalfLifeDataset
+::: beignet.datasets.PDB70Dataset
+::: beignet.datasets.PDBTrajectoryDataset
+::: beignet.datasets.PDBbindDataset
+::: beignet.datasets.ParquetDataset
+::: beignet.datasets.QM7Dataset
+::: beignet.datasets.QM7bDataset
+::: beignet.datasets.QM8Dataset
+::: beignet.datasets.QM9Dataset
+::: beignet.datasets.REALDatabaseDataset
+::: beignet.datasets.RandomEulerAngleDataset
+::: beignet.datasets.RandomQuaternionDataset
+::: beignet.datasets.RandomRotationDataset
+::: beignet.datasets.RandomRotationMatrixDataset
+::: beignet.datasets.RandomRotationVectorDataset
+::: beignet.datasets.SAbDabDataset
+::: beignet.datasets.SKEMPIDataset
+::: beignet.datasets.SequenceDataset
+::: beignet.datasets.SizedSequenceDataset
+::: beignet.datasets.SwissProtDataset
+::: beignet.datasets.TDCDataset
+::: beignet.datasets.TherapeuticAntibodyProfilerDataset
+::: beignet.datasets.Tox21Dataset
+::: beignet.datasets.ToxCastDataset
+::: beignet.datasets.TrEMBLDataset
+::: beignet.datasets.TrajectoryDataset
+::: beignet.datasets.USPTODataset
+::: beignet.datasets.USPTOReactionProductDataset
+::: beignet.datasets.UniProtDataset
+::: beignet.datasets.UniRef100Dataset
+::: beignet.datasets.UniRef50Dataset
+::: beignet.datasets.UniRef90Dataset
+::: beignet.datasets.VeithCytochromeP4501A2InhibitionDataset
+::: beignet.datasets.VeithCytochromeP4502C19InhibitionDataset
+::: beignet.datasets.VeithCytochromeP4502C9InhibitionDataset
+::: beignet.datasets.VeithCytochromeP4502D6InhibitionDataset
+::: beignet.datasets.VeithCytochromeP4503A4InhibitionDataset
+::: beignet.datasets.WangEffectivePermeabilityDataset
+::: beignet.datasets.ZINCDataset
+::: beignet.datasets.ZhuAcuteToxicityLD50Dataset
diff --git a/docs/reference/beignet.io.md b/docs/reference/beignet.io.md
@@ -1,3 +1 @@
-# beignet.io
-
 ::: beignet.io.ThreadSafeFile
diff --git a/docs/reference/datasets/antibody-affinity.md b/docs/reference/datasets/antibody-affinity.md
@@ -0,0 +1 @@
+::: beignet.datasets.SAbDabDataset
diff --git a/docs/reference/datasets/catalyst.md b/docs/reference/datasets/catalyst.md
diff --git a/docs/reference/datasets/datasets.md b/docs/reference/datasets/datasets.md
diff --git a/docs/reference/datasets/drug-interaction.md b/docs/reference/datasets/drug-interaction.md
@@ -0,0 +1,3 @@
+::: beignet.datasets.BindingDBKdDataset
+::: beignet.datasets.DAVISDataset
+::: beignet.datasets.KIBADataset
diff --git a/docs/reference/datasets/drug-intolerance.md b/docs/reference/datasets/drug-intolerance.md
diff --git a/docs/reference/datasets/epitope.md b/docs/reference/datasets/epitope.md
diff --git a/docs/reference/datasets/gene-disease-association.md b/docs/reference/datasets/gene-disease-association.md
@@ -0,0 +1 @@
+::: beignet.datasets.DisGeNETDataset
diff --git a/docs/reference/datasets/general-purpose.md b/docs/reference/datasets/general-purpose.md
@@ -0,0 +1,8 @@
+::: beignet.datasets.ATOM3DDataset
+::: beignet.datasets.DataFrameDataset
+::: beignet.datasets.FASTADataset
+::: beignet.datasets.LMDBDataset
+::: beignet.datasets.ParquetDataset
+::: beignet.datasets.SequenceDataset
+::: beignet.datasets.SizedSequenceDataset
+::: beignet.datasets.TDCDataset
diff --git a/docs/reference/datasets/geometry/transformations.md b/docs/reference/datasets/geometry/transformations.md
@@ -1,3 +1,4 @@
+::: beignet.datasets.RandomRotationDataset
 ::: beignet.datasets.RandomEulerAngleDataset
 ::: beignet.datasets.RandomQuaternionDataset
 ::: beignet.datasets.RandomRotationMatrixDataset

diff --git a/docs/reference/datasets/molecule.md b/docs/reference/datasets/molecule.md
@@ -0,0 +1,3 @@
+::: beignet.datasets.ChEMBLDataset
+::: beignet.datasets.MOSESDataset
+::: beignet.datasets.ZINCDataset
diff --git a/docs/reference/datasets/pharmacokinetic-property/absorption.md b/docs/reference/datasets/pharmacokinetic-property/absorption.md
@@ -0,0 +1,7 @@
+::: beignet.datasets.AqSolDBSolubilityDataset
+::: beignet.datasets.AstraZenecaLipophilicityDataset
+::: beignet.datasets.BroccatelliPGlycoproteinInhibitionDataset
+::: beignet.datasets.HouHumanIntestinalAbsorptionDataset
+::: beignet.datasets.MaBioavailabilityDataset
+::: beignet.datasets.NCATSPAMPAPermeabilityDataset
+::: beignet.datasets.WangEffectivePermeabilityDataset
diff --git a/docs/reference/datasets/pharmacokinetic-property/distribution.md b/docs/reference/datasets/pharmacokinetic-property/distribution.md
@@ -0,0 +1,3 @@
+::: beignet.datasets.AstraZenecaPlasmaProteinBindingRateDataset
+::: beignet.datasets.LombardoVolumeOfDistributionAtSteadyStateDataset
+::: beignet.datasets.MartinsBloodBrainBarrierDataset
diff --git a/docs/reference/datasets/pharmacokinetic-property/excretion.md b/docs/reference/datasets/pharmacokinetic-property/excretion.md
@@ -0,0 +1,2 @@
+::: beignet.datasets.AstraZenecaClearanceDataset
+::: beignet.datasets.ObachHalfLifeDataset
diff --git a/docs/reference/datasets/pharmacokinetic-property/metabolism.md b/docs/reference/datasets/pharmacokinetic-property/metabolism.md
@@ -0,0 +1,8 @@
+::: beignet.datasets.CarbonMangelsCytochromeP4502C9SubstrateDataset
+::: beignet.datasets.CarbonMangelsCytochromeP4502D6SubstrateDataset
+::: beignet.datasets.CarbonMangelsCytochromeP4503A4SubstrateDataset
+::: beignet.datasets.VeithCytochromeP4501A2InhibitionDataset
+::: beignet.datasets.VeithCytochromeP4502C19InhibitionDataset
+::: beignet.datasets.VeithCytochromeP4502C9InhibitionDataset
+::: beignet.datasets.VeithCytochromeP4502D6InhibitionDataset
+::: beignet.datasets.VeithCytochromeP4503A4InhibitionDataset
diff --git a/docs/reference/datasets/protein-protein-interaction.md b/docs/reference/datasets/protein-protein-interaction.md
@@ -0,0 +1 @@
+::: beignet.datasets.HuRIDataset
diff --git a/docs/reference/datasets/protein/protein-sequence.md b/docs/reference/datasets/protein/protein-sequence.md
@@ -0,0 +1,6 @@
+::: beignet.datasets.UniProtDataset
+::: beignet.datasets.UniRef50Dataset
+::: beignet.datasets.UniRef90Dataset
+::: beignet.datasets.UniRef100Dataset
+::: beignet.datasets.SwissProtDataset
+::: beignet.datasets.TrEMBLDataset
diff --git a/docs/reference/datasets/protein/protein-structure.md b/docs/reference/datasets/protein/protein-structure.md
diff --git a/docs/reference/datasets/quantum-mechanics.md b/docs/reference/datasets/quantum-mechanics.md
@@ -0,0 +1,4 @@
+::: beignet.datasets.QM7Dataset
+::: beignet.datasets.QM7bDataset
+::: beignet.datasets.QM8Dataset
+::: beignet.datasets.QM9Dataset
diff --git a/docs/reference/datasets/reaction.md b/docs/reference/datasets/reaction.md
@@ -0,0 +1,2 @@
+::: beignet.datasets.USPTOReactionProductDataset
+::: beignet.datasets.USPTODataset
diff --git a/docs/reference/datasets/retrosynthesis.md b/docs/reference/datasets/retrosynthesis.md
diff --git a/docs/reference/datasets/sequences.md b/docs/reference/datasets/sequences.md
@@ -1,3 +1,14 @@
-::: beignet.datasets.FASTADataset
-::: beignet.datasets.SequenceDataset
-::: beignet.datasets.SizedSequenceDataset
+::: beignet.datasets.ATOM3DMSPDataset
+::: beignet.datasets.ATOM3DPPIDataset
+::: beignet.datasets.ATOM3DPSRDataset
+::: beignet.datasets.ATOM3DRESDataset
+::: beignet.datasets.ATOM3DRSRDataset
+::: beignet.datasets.ATOM3DSMPDataset
+::: beignet.datasets.DrugCombDataset
+::: beignet.datasets.FreeSolvDataset
+::: beignet.datasets.GDSC1Dataset
+::: beignet.datasets.GDSC2Dataset
+::: beignet.datasets.PDB70Dataset
+::: beignet.datasets.PDBbindDataset
+::: beignet.datasets.REALDatabaseDataset
+::: beignet.datasets.SKEMPIDataset
diff --git a/docs/reference/datasets/toxicity.md b/docs/reference/datasets/toxicity.md
@@ -0,0 +1,4 @@
+::: beignet.datasets.ClinToxDataset
+::: beignet.datasets.Tox21Dataset
+::: beignet.datasets.ToxCastDataset
+::: beignet.datasets.ZhuAcuteToxicityLD50Dataset
diff --git a/mkdocs.yml b/mkdocs.yml
@@ -34,19 +34,15 @@ nav:
               - "Probabilists’ Hermite polynomial": "reference/operators/special-functions/orthogonal-polynomials/probabilists-hermite-polynomial.md"
               - "Laguerre polynomial": "reference/operators/special-functions/orthogonal-polynomials/laguerre-polynomial.md"
               - "Legendre polynomial": "reference/operators/special-functions/orthogonal-polynomials/legendre-polynomial.md"
-      - "Thermodynamics": "reference/reference/operators/thermodynamics.md"
-    - "beignet.datasets":
-        - "Geometry":
-            - "Transformations": "reference/datasets/geometry/transformations.md"
-        - "Protein sequences": "reference/datasets/protein-sequences.md"
-        - "Sequences": "reference/datasets/sequences.md"
+      - "Thermodynamics": "reference/operators/thermodynamics.md"
+    - "beignet.datasets": "reference/beignet.datasets.md"
     - "beignet.features":
         - "General-purpose": "reference/features/general-purpose.md"
         - "Geometry":
             - "Transformations": "reference/features/geometry/transformations.md"
     - "beignet.func":
         - "Molecular dynamics": "reference/func/molecular-dynamics.md"
-    - "beignet.io": "reference/operators.io.md"
+    - "beignet.io": "reference/beignet.io.md"
     - "beignet.transforms":
         - "General-purpose": "reference/transforms/general-purpose.md"
 plugins:

diff --git a/pyproject.toml b/pyproject.toml
@@ -22,7 +22,13 @@ requires-python = ">=3.10"
 
 [project.optional-dependencies]
 all = [
-    "beignet[docs,mdtraj,test]",
+    "beignet[datasets,docs,mdtraj,test]",
+]
+datasets = [
+    "biopython",
+    "lmdb",
+    "pandas",
+    "pooch",
 ]
 docs = [
     "mkdocs-material",
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,2 @@
		::: beignet.datasets.AstraZenecaClearanceDataset
		::: beignet.datasets.ObachHalfLifeDataset
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,2 @@
		::: beignet.datasets.USPTOReactionProductDataset
		::: beignet.datasets.USPTODataset