Merge pull request #107 from volkamerlab/klifs-drugs

Add KLIFS drugs data

opencadd/data/klifs_fields.csv

@@ -99,3 +99,11 @@ coordinates,residue.klifs_id,Int32,,,
 coordinates,residue.klifs_region_id,string,,,
 coordinates,residue.klifs_region,string,,,
 coordinates,residue.klifs_color,string,,,
+drugs,drug.inn,string,INN,,
+drugs,drug.brand_name,string,Brand name,,
+drugs,drug.synonym,string,Synonyms,,
+drugs,drug.phase,string,Phase,,
+drugs,drug.approval_year,string,Approval,,
+drugs,drug.smiles,string,SMILES,,
+drugs,ligand.chembl_id,string,ChEMBL,,
+drugs,ligand.expo_id,string,PDB,,

opencadd/databases/klifs/core.py

@@ -138,19 +138,25 @@ def _standardize_column_values(dataframe):
         local data is already performed upon session initialization.
         """
 
+        # TODO Use None instead of "-"; but may affect downstream pipelines that use "-" already
         if "structure.alternate_model" in dataframe.columns:
-            # Remote
             dataframe["structure.alternate_model"].replace("", "-", inplace=True)
         if "ligand.expo_id" in dataframe.columns:
-            # Remote
             dataframe["ligand.expo_id"].replace(0, "-", inplace=True)
         if "ligand_allosteric.expo_id" in dataframe.columns:
-            # Remote
             dataframe["ligand_allosteric.expo_id"].replace(0, "-", inplace=True)
         if "structure.resolution" in dataframe.columns:
-            # Remote
             dataframe["structure.resolution"].replace(0, np.nan, inplace=True)
 
+        if "drug.brand_name" in dataframe.columns:
+            dataframe["drug.brand_name"] = dataframe["drug.brand_name"].apply(
+                lambda x: x.split(";") if x != "" else []
+            )
+        if "drug.synonyms" in dataframe.columns:
+            dataframe["drug.synonyms"] = dataframe["drug.synonyms"].apply(
+                lambda x: x.split("\t") if x != "" else []
+            )
+
         return dataframe
 
     def _standardize_dataframe(self, dataframe, columns, columns_mapping=None):
@@ -1046,10 +1052,13 @@ class InteractionsProvider(BaseProvider):
     """
     Class for interactions requests.
 
-    Methods
-    -------
+    Properties
+    ----------
     interaction_types()
         Get all available interaction types.
+
+    Methods
+    -------
     all_interactions()
         Get all available interaction fingerprints.
     by_structure_klifs_id(structure_klifs_ids)
@@ -1428,3 +1437,62 @@ def _raise_invalid_extension(extension):
         extensions = ["pdb", "mol2"]
         if extension not in extensions:
             raise ValueError(f"Invalid extension. Select from: {', '.join(extensions)}")
+
+
+class DrugsProvider(BaseProvider):
+    """
+    Class for drugs requests.
+
+    From the KLIFS Swagger API:
+    https://dev.klifs.net/swagger_v2/#/Ligands/get_drug_list
+    > The drug list endpoint returns a list of all annotated kinase ligands that are either
+    > approved or are/have been in clinical trials.
+    > This information is primarily powered by the PKIDB and complemented with KLIFS curation and
+    > annotation + manually curated data from other sources (e.g. approved INNs).
+
+    Methods
+    -------
+    all_drugs()
+        Get all available drugs.
+
+    Notes
+    -----
+    Class methods all return a pandas.DataFrame of drugs (rows) with the following attributes
+    (columns):
+
+    drug.inn : string
+        International nonproprietary name.
+    drug.brand_name : list of string
+        Brand name(s).
+    drug.synonym : list of string
+        Synonym(s).
+    drug.phase : string
+        Current clinical phase of the drug.
+    drug.approval_year : string
+        Year of FDA-approval.
+        If approval by another institution, syntax as follows, example: "2017 (EMA)".
+    drug.smiles : string
+        SMILES string of drug.
+        TODO: "ligand.smiles" would be more consistent with Ligand class, howover it is not
+        garanteed that SMILES will be the same for the same ligand, thus use "drug.smiles".
+    ligand.chembl_id : string
+        Ligand ChEMBL ID.
+    ligand.expo_id : string
+        Ligand expo ID.
+    """
+
+    def all_drugs(self):
+        """
+        Get all available drugs.
+
+        Returns
+        -------
+        pandas.DataFrame
+            drugs (rows) with the columns as defined in the class docstring.
+
+        Raises
+        ------
+        ValueError
+            If DataFrame is empty.
+        """
+        raise NotImplementedError("Implement in your subclass!")

opencadd/databases/klifs/local.py

@@ -19,6 +19,7 @@
     InteractionsProvider,
     PocketsProvider,
     CoordinatesProvider,
+    DrugsProvider,
 )
 from .schema import (
     FIELDS,
@@ -879,3 +880,17 @@ def _add_residue_klifs_ids(self, dataframe, filepath):
         dataframe = dataframe.merge(pocket_dataframe, on="residue.id", how="left")
 
         return dataframe
+
+
+class Drugs(LocalInitializer, DrugsProvider):
+    """
+    Extends DrugsProvider to provide remote drug requests.
+    Refer to DrugsProvider documentation for more information:
+    opencadd.databases.klifs.core.DrugsProvider
+    """
+
+    def all_drugs(self):
+
+        raise NotImplementedError(
+            "Information on drugs is not available locally! Please use a remote session."
+        )

opencadd/databases/klifs/remote.py

@@ -19,6 +19,7 @@
     InteractionsProvider,
     PocketsProvider,
     CoordinatesProvider,
+    DrugsProvider,
 )
 from .schema import FIELDS
 from .utils import metadata_to_filepath, silence_logging
@@ -869,3 +870,23 @@ def _add_residue_klifs_ids(self, dataframe, structure_klifs_id):
         dataframe = dataframe.astype({"residue.klifs_id": "Int64"})
 
         return dataframe
+
+
+class Drugs(RemoteInitializer, DrugsProvider):
+    """
+    Extends DrugsProvider to provide remote drug requests.
+    Refer to DrugsProvider documentation for more information:
+    opencadd.databases.klifs.core.DrugsProvider
+    """
+
+    def all_drugs(self):
+
+        # Use KLIFS API
+        result = self._client.Ligands.get_drug_list().response().result
+        # Convert list of ABC objects to DataFrame
+        drugs = self._abc_to_dataframe(result)
+        # Standardize DataFrame
+        drugs = self._standardize_dataframe(
+            drugs, FIELDS.oc_name_to_type("drugs"), FIELDS.remote_to_oc_names("drugs")
+        )
+        return drugs

opencadd/databases/klifs/session.py

@@ -37,6 +37,8 @@ class Session:
         Interactions object for interaction requests.
     coordinates : None or opencadd.databases.klifs.remote.Coordinates/local.Coordinates
         Coordinates object for coordinates requests.
+    drugs : None or opencadd.databases.klifs.remote.Drugs
+        Coordinates object for drugs requests.
     """
 
     def __init__(self):
@@ -56,6 +58,7 @@ def __init__(self):
         self.interactions = None
         self.pockets = None
         self.coordinates = None
+        self.drugs = None
 
     @classmethod
     def from_local(cls, path_to_klifs_download, path_to_klifs_metadata=None):
@@ -165,3 +168,8 @@ def _set_attributes(self, backend, path_to_klifs_download=None, database=None, c
             database=database,
             path_to_klifs_download=path_to_klifs_download,
         )
+        self.drugs = backend.Drugs(
+            client=client,
+            database=database,
+            path_to_klifs_download=path_to_klifs_download,
+        )

opencadd/databases/klifs/utils.py

@@ -66,7 +66,7 @@ def metadata_to_filepath(
     structure = f"{structure_pdb}{f'_alt{structure_alternate_model}' if bool(structure_alternate_model) else ''}{f'_chain{structure_chain}' if bool(structure_chain) else ''}"
 
     # FIXME: The PDB download for ligands in KLIFS is named "klifs_ligand.pdb"
-    # instead of "ligand.expo_id". For the time being (until KLIFS maybe streamlines the file name
+    # instead of "ligand.pdb". For the time being (until KLIFS maybe streamlines the file name
     # with all the other file names), rename the file here.
     if entity == "ligand" and extension == "pdb":
         entity = "klifs_ligand"

opencadd/tests/databases/test_klifs_local_remote.py

@@ -225,6 +225,17 @@ def test_all_bioactivities(self):
 
         check_dataframe(result_remote, FIELDS.oc_name_to_type("bioactivities"))
 
+    def test_all_drugs(self):
+        """
+        Test request result for all drugs.
+        """
+
+        result_remote = REMOTE.drugs.all_drugs()
+        check_dataframe(result_remote, FIELDS.oc_name_to_type("drugs"))
+
+        with pytest.raises(NotImplementedError):
+            LOCAL.drugs.all_drugs()
+
 
 class TestsFromKinaseIds:
     """