Update drugcentral.py

src/pyobo/sources/drugcentral.py

@@ -3,12 +3,15 @@
 """Get DrugCentral as OBO."""
 
 import logging
-from typing import Iterable
+from collections import defaultdict
+from contextlib import closing
+from typing import DefaultDict, Iterable, List
 
-import pandas as pd
+import bioregistry
+import psycopg2
+from tqdm.auto import tqdm
 
-from pyobo.struct import Obo, Reference, Term
-from pyobo.utils.path import ensure_df
+from pyobo.struct import Obo, Reference, Synonym, Term
 
 __all__ = [
     "DrugCentralGetter",
@@ -18,6 +21,13 @@
 
 PREFIX = "drugcentral"
 
+HOST = "unmtid-dbs.net"
+PORT = 5433
+USER = "drugman"
+PASSWORD = "dosage"
+DBNAME = "drugcentral"
+PARAMS = dict(dbname=DBNAME, user=USER, password=PASSWORD, host=HOST, port=PORT)
+
 
 class DrugCentralGetter(Obo):
     """An ontology representation of the DrugCentral database."""
@@ -26,27 +36,59 @@ class DrugCentralGetter(Obo):
 
     def iter_terms(self, force: bool = False) -> Iterable[Term]:
         """Iterate over terms in the ontology."""
-        return iter_terms(version=self._version_or_raise, force=force)
+        return iter_terms()
 
 
 def get_obo(force: bool = False) -> Obo:
     """Get DrugCentral OBO."""
     return DrugCentralGetter(force=force)
 
 
-def iter_terms(version: str, force: bool = False) -> Iterable[Term]:
+def iter_terms() -> Iterable[Term]:
     """Iterate over DrugCentral terms."""
-    url = f"https://unmtid-shinyapps.net/download/DrugCentral/{version}/structures.smiles.tsv"
-    df = ensure_df(PREFIX, url=url, version=version, force=force)
-    for smiles, inchi, inchi_key, drugcentral_id, drugcentral_name, cas in df.values:
-        if pd.isna(smiles) or pd.isna(inchi) or pd.isna(inchi_key):
-            logger.warning("missing data for drugcentral:%s", drugcentral_id)
-            continue
-        term = Term.from_triple(prefix=PREFIX, identifier=drugcentral_id, name=drugcentral_name)
+    with closing(psycopg2.connect(**PARAMS)) as conn:
+        with closing(conn.cursor()) as cur:
+            cur.execute(
+                "SELECT cd_id, name, cas_reg_no, mrdef, inchi, smiles, inchikey FROM public.structures"
+            )
+            structures = cur.fetchall()
+
+        with closing(conn.cursor()) as cur:
+            cur.execute("SELECT struct_id, id_type, identifier FROM public.identifier")
+            rows = cur.fetchall()
+            xrefs: DefaultDict[str, List[Reference]] = defaultdict(list)
+            for drugcentral_id, prefix, identifier in tqdm(
+                rows, unit_scale=True, desc="loading xrefs"
+            ):
+                if not identifier or not prefix:
+                    continue
+                if prefix == "ChEMBL_ID":
+                    prefix = "chembl.compound"
+                xref_prefix_norm = bioregistry.normalize_prefix(prefix)
+                if xref_prefix_norm is None:
+                    tqdm.write(f"did not normalize {prefix}:{identifier}")
+                    continue
+                xrefs[str(drugcentral_id)].append(
+                    Reference(prefix=xref_prefix_norm, identifier=identifier)
+                )
+        with closing(conn.cursor()) as cur:
+            cur.execute("SELECT id, name FROM public.synonyms")
+            synonyms: DefaultDict[str, List[Synonym]] = defaultdict(list)
+            for drugcentral_id, synonym in cur.fetchall():
+                synonyms[str(drugcentral_id)].append(Synonym(name=synonym))
+
+    for drugcentral_id, name, cas, definition, inchi, smiles, inchi_key in structures:
+        drugcentral_id = str(drugcentral_id)
+        term = Term(
+            reference=Reference(prefix=PREFIX, identifier=drugcentral_id, name=name),
+            definition=definition,
+            synonyms=synonyms.get(drugcentral_id, []),
+            xrefs=xrefs.get(drugcentral_id, []),
+        )
         term.append_xref(Reference(prefix="inchikey", identifier=inchi_key))
         term.append_property("smiles", smiles)
         term.append_property("inchi", inchi)
-        if pd.notna(cas):
+        if cas:
             term.append_xref(Reference(prefix="cas", identifier=cas))
         yield term