Update genesis/api_clients/bioportal_api.py

genesis/api_clients/bioportal_api.py

@@ -1,73 +1,89 @@
 # genesis/api_clients/bioportal_api.py
 import os
 import requests
-from typing import List, Dict, Optional
+from typing import Dict, List, Optional
 
-BIOPORTAL_API_KEY = os.getenv("BIOPORTAL_API_KEY")  # From your Hugging Face secrets or .env
+BIOPORTAL_API_KEY = os.getenv("BIOPORTAL_API_KEY")
 BIOPORTAL_BASE = "https://data.bioontology.org"
 
-HEADERS = {
-    "Authorization": f"apikey token={BIOPORTAL_API_KEY}"
-}
+if not BIOPORTAL_API_KEY:
+    raise EnvironmentError("Missing BIOPORTAL_API_KEY in environment variables.")
 
-def search_ontology(term: str, ontology: Optional[str] = None, max_results: int = 10) -> List[Dict]:
+# -------------------------
+# CORE HELPERS
+# -------------------------
+def _auth_params(params: Dict) -> Dict:
+    """Attach BioPortal API key."""
+    params["apikey"] = BIOPORTAL_API_KEY
+    return params
+
+def search_term(term: str, ontologies: Optional[List[str]] = None, max_results: int = 10) -> List[Dict]:
     """
-    Search BioPortal for a given term, optionally within a specific ontology.
+    Search BioPortal for ontology terms.
+    `ontologies` is optional list like ["GO", "CHEBI", "NCBITAXON"]
     """
-    params = {
+    params = _auth_params({
         "q": term,
         "pagesize": max_results
-    }
-    if ontology:
-        params["ontology"] = ontology
-    
-    r = requests.get(f"{BIOPORTAL_BASE}/search", params=params, headers=HEADERS)
+    })
+    if ontologies:
+        params["ontologies"] = ",".join(ontologies)
+
+    r = requests.get(f"{BIOPORTAL_BASE}/search", params=params)
     r.raise_for_status()
     data = r.json()
-    
+
     results = []
-    for item in data.get("collection", []):
+    for collection in data.get("collection", []):
         results.append({
-            "label": item.get("prefLabel"),
-            "definition": item.get("definition"),
-            "synonyms": item.get("synonym", []),
-            "ontology": item.get("links", {}).get("ontology"),
-            "iri": item.get("@id"),
-            "cui": item.get("cui", None),
-            "semantic_types": item.get("semanticType", [])
+            "prefLabel": collection.get("prefLabel"),
+            "ontology": collection.get("links", {}).get("ontology"),
+            "definition": collection.get("definition"),
+            "cui": collection.get("cui"),
+            "uri": collection.get("@id")
         })
     return results
 
-def get_term_details(ontology_acronym: str, term_id: str) -> Dict:
+def get_term_details(term_uri: str) -> Dict:
     """
-    Get detailed information about a specific term in an ontology.
+    Get detailed metadata for a term given its URI.
     """
-    encoded_term_id = term_id.replace(":", "%3A").replace("/", "%2F")
-    r = requests.get(
-        f"{BIOPORTAL_BASE}/ontologies/{ontology_acronym}/classes/{encoded_term_id}",
-        headers=HEADERS
-    )
+    params = _auth_params({})
+    r = requests.get(term_uri, params=params)
     r.raise_for_status()
     return r.json()
 
-def map_to_ontologies(term: str) -> List[Dict]:
+def get_mappings(term_uri: str) -> List[Dict]:
     """
-    Search multiple ontologies for mappings of a term.
+    Get cross-ontology mappings for a term.
+    Useful for linking to UMLS, NCBI, ChEMBL.
     """
-    params = {
-        "q": term,
-        "include": "mappings"
-    }
-    r = requests.get(f"{BIOPORTAL_BASE}/search", params=params, headers=HEADERS)
+    params = _auth_params({})
+    r = requests.get(f"{term_uri}/mappings", params=params)
     r.raise_for_status()
     data = r.json()
-    
+
     mappings = []
-    for item in data.get("collection", []):
-        for mapping in item.get("mappings", []):
-            mappings.append({
-                "source": mapping.get("source"),
-                "target": mapping.get("target"),
-                "relation": mapping.get("relation")
-            })
+    for m in data:
+        mappings.append({
+            "source": m.get("classes", [{}])[0].get("links", {}).get("ontology"),
+            "target": m.get("classes", [{}])[-1].get("links", {}).get("ontology"),
+            "target_uri": m.get("classes", [{}])[-1].get("@id"),
+            "target_label": m.get("classes", [{}])[-1].get("prefLabel")
+        })
     return mappings
+
+def find_related_terms(term: str, ontologies: Optional[List[str]] = None) -> List[Dict]:
+    """
+    Find related ontology terms using BioPortal search + mappings.
+    """
+    results = search_term(term, ontologies=ontologies)
+    related = []
+    for r in results:
+        mappings = get_mappings(r["uri"])
+        related.append({
+            "term": r["prefLabel"],
+            "uri": r["uri"],
+            "mappings": mappings
+        })
+    return related