Update genesis/api_clients/bioportal_api.py

genesis/api_clients/bioportal_api.py

@@ -1,97 +1,73 @@
 # genesis/api_clients/bioportal_api.py
-import requests
-from typing import Dict, List, Optional
 import os
+import requests
+from typing import List, Dict, Optional
+
+# BioPortal API
+BIOPORTAL_BASE = "https://data.bioontology.org"
+BIOPORTAL_API_KEY = os.getenv("BIOPORTAL_API_KEY")  # You must add this to Hugging Face secrets or .env
 
-BIOPORTAL_API_KEY = os.getenv("BIOPORTAL_API_KEY")  # Stored in Hugging Face secrets
-BIOPORTAL_BASE_URL = "https://data.bioontology.org"
+if not BIOPORTAL_API_KEY:
+    raise EnvironmentError("BIOPORTAL_API_KEY is missing. Add it to your environment variables or Hugging Face secrets.")
 
-def bioportal_search(term: str, ontologies: Optional[List[str]] = None, exact_match: bool = False, max_results: int = 20) -> List[Dict]:
+HEADERS = {
+    "Authorization": f"apikey token={BIOPORTAL_API_KEY}"
+}
+
+def search_terms(query: str, ontologies: Optional[List[str]] = None, max_results: int = 20) -> List[Dict]:
     """
     Search BioPortal for ontology terms.
-    ontologies: list of ontology acronyms to restrict search (e.g., ["MESH", "SNOMEDCT", "GO"])
-    exact_match: if True, only return exact matches
+    Optionally limit search to specific ontologies like 'GO', 'CHEBI', 'DOID', etc.
     """
-    if not BIOPORTAL_API_KEY:
-        raise ValueError("BIOPORTAL_API_KEY not set in environment.")
-
     params = {
-        "apikey": BIOPORTAL_API_KEY,
-        "q": term,
-        "pagesize": max_results,
-        "require_exact_match": "true" if exact_match else "false"
+        "q": query,
+        "pagesize": max_results
     }
-
     if ontologies:
         params["ontologies"] = ",".join(ontologies)
-
-    r = requests.get(f"{BIOPORTAL_BASE_URL}/search", params=params)
+    
+    r = requests.get(f"{BIOPORTAL_BASE}/search", headers=HEADERS, params=params)
     r.raise_for_status()
     data = r.json()
-
+    
     results = []
     for item in data.get("collection", []):
         results.append({
-            "prefLabel": item.get("prefLabel", ""),
-            "definition": (item.get("definition") or ["No definition"])[0],
+            "prefLabel": item.get("prefLabel"),
+            "definition": item.get("definition", []),
             "synonyms": item.get("synonym", []),
             "ontology": item.get("links", {}).get("ontology"),
-            "iri": item.get("@id"),
+            "uri": item.get("@id"),
             "cui": item.get("cui", []),
             "semanticType": item.get("semanticType", []),
-            "parents": item.get("links", {}).get("parents"),
-            "children": item.get("links", {}).get("children")
         })
-
     return results
 
 
-def bioportal_autocomplete(prefix: str, ontologies: Optional[List[str]] = None, max_results: int = 10) -> List[str]:
+def get_term_details(term_uri: str) -> Dict:
     """
-    Autocomplete search for ontology terms by prefix.
+    Fetch full details for a given ontology term URI.
     """
-    if not BIOPORTAL_API_KEY:
-        raise ValueError("BIOPORTAL_API_KEY not set in environment.")
-
-    params = {
-        "apikey": BIOPORTAL_API_KEY,
-        "q": prefix,
-        "pagesize": max_results
-    }
-
-    if ontologies:
-        params["ontologies"] = ",".join(ontologies)
-
-    r = requests.get(f"{BIOPORTAL_BASE_URL}/search", params=params)
+    encoded_uri = requests.utils.quote(term_uri, safe="")
+    r = requests.get(f"{BIOPORTAL_BASE}/ontologies/{encoded_uri}", headers=HEADERS)
     r.raise_for_status()
-    data = r.json()
-
-    return [item.get("prefLabel", "") for item in data.get("collection", []) if item.get("prefLabel")]
+    return r.json()
 
 
-def get_term_details(ontology_acronym: str, term_id: str) -> Dict:
+def find_related_terms(term_uri: str) -> List[Dict]:
     """
-    Retrieve full details for a specific term in an ontology.
+    Find related terms for a given ontology concept.
     """
-    if not BIOPORTAL_API_KEY:
-        raise ValueError("BIOPORTAL_API_KEY not set in environment.")
-
-    r = requests.get(f"{BIOPORTAL_BASE_URL}/ontologies/{ontology_acronym}/classes/{term_id}", params={"apikey": BIOPORTAL_API_KEY})
+    encoded_uri = requests.utils.quote(term_uri, safe="")
+    r = requests.get(f"{BIOPORTAL_BASE}/ontologies/{encoded_uri}/related", headers=HEADERS)
     r.raise_for_status()
     return r.json()
 
 
-def map_to_umls(term: str) -> List[Dict]:
+def map_to_ontology(entity_name: str, preferred_ontologies: Optional[List[str]] = None) -> Optional[Dict]:
     """
-    Map a biomedical term to UMLS concepts via BioPortal mappings.
+    Try to map a free-text biomedical entity name to an ontology concept.
+    Useful for linking PubMed entities, ChEMBL molecules, clinical trials to ontology terms.
     """
-    mappings = bioportal_search(term)
-    umls_concepts = []
-    for m in mappings:
-        if "cui" in m and m["cui"]:
-            umls_concepts.append({
-                "term": m["prefLabel"],
-                "cui": m["cui"],
-                "semanticType": m["semanticType"]
-            })
-    return umls_concepts
+    matches = search_terms(entity_name, ontologies=preferred_ontologies, max_results=1)
+    return matches[0] if matches else None