Update genesis/api_clients/bioportal_api.py

genesis/api_clients/bioportal_api.py

@@ -1,103 +1,88 @@
 # genesis/api_clients/bioportal_api.py
 import os
 import requests
-from typing import Dict, List, Optional
+from typing import List, Dict
 
+BIOPORTAL_API_KEY = os.getenv("BIOPORTAL_API_KEY")  # Set in Hugging Face secrets or .env
 BIOPORTAL_BASE = "https://data.bioontology.org"
-BIOPORTAL_API_KEY = os.getenv("BIOPORTAL_API_KEY")  # Saved in Hugging Face / .env
 
 # -------------------------
-# Core Request Helper
+# Search Ontology Terms
 # -------------------------
-def bioportal_request(endpoint: str, params: Dict) -> Dict:
+def search_ontology(term: str, max_results: int = 5) -> List[Dict]:
     """
-    Helper for BioPortal API requests.
+    Search BioPortal for matching ontology terms across multiple vocabularies.
     """
     if not BIOPORTAL_API_KEY:
-        raise ValueError("BIOPORTAL_API_KEY not set in environment variables")
-    
-    headers = {"Authorization": f"apikey token={BIOPORTAL_API_KEY}"}
-    r = requests.get(f"{BIOPORTAL_BASE}{endpoint}", headers=headers, params=params)
+        raise ValueError("BioPortal API key not set in environment variables.")
+
+    params = {
+        "q": term,
+        "pagesize": max_results,
+        "apikey": BIOPORTAL_API_KEY
+    }
+    r = requests.get(f"{BIOPORTAL_BASE}/search", params=params)
     r.raise_for_status()
-    return r.json()
 
-# -------------------------
-# Search Ontology Terms
-# -------------------------
-def search_terms(query: str, ontology: Optional[str] = None, max_results: int = 10) -> List[Dict]:
-    """
-    Search ontology terms across all ontologies or within a specific ontology.
-    ontology: Optional ontology acronym (e.g., MESH, GO, SNOMEDCT)
-    """
-    params = {"q": query, "pagesize": max_results}
-    if ontology:
-        params["ontology"] = ontology
-    data = bioportal_request("/search", params)
-    return data.get("collection", [])
+    results = r.json().get("collection", [])
+    terms = []
+    for item in results:
+        terms.append({
+            "prefLabel": item.get("prefLabel"),
+            "ontology": item.get("links", {}).get("ontology"),
+            "definition": item.get("definition", [""])[0] if item.get("definition") else "",
+            "iri": item.get("@id"),
+            "synonyms": item.get("synonym", []),
+            "cui": item.get("cui", []),  # UMLS Concept Unique Identifiers if available
+        })
+
+    return terms
 
 # -------------------------
 # Get Term Details
 # -------------------------
-def get_term_details(ontology: str, term_id: str) -> Dict:
+def get_term_details(ontology_acronym: str, term_id: str) -> Dict:
     """
-    Retrieve details for a specific ontology term.
+    Get full details for a specific ontology term.
     """
-    return bioportal_request(f"/ontologies/{ontology}/classes/{term_id}", {})
+    params = {
+        "apikey": BIOPORTAL_API_KEY
+    }
+    r = requests.get(f"{BIOPORTAL_BASE}/ontologies/{ontology_acronym}/classes/{term_id}", params=params)
+    r.raise_for_status()
+    return r.json()
 
 # -------------------------
-# Mapping & Crosslinks
+# Map Term to UMLS CUI
 # -------------------------
-def get_term_mappings(term_id: str) -> List[Dict]:
-    """
-    Retrieve mappings for an ontology term to other ontologies.
+def map_to_umls(term: str) -> List[str]:
     """
-    data = bioportal_request(f"/mappings/{term_id}", {})
-    return data.get("collection", [])
-
-def search_and_map(query: str, ontology: Optional[str] = None) -> List[Dict]:
-    """
-    Search for a term and retrieve mappings.
+    Map a given term to UMLS Concept Unique Identifiers (CUIs).
     """
-    terms = search_terms(query, ontology)
-    mapped_results = []
-    for t in terms:
-        mappings = get_term_mappings(t["@id"])
-        mapped_results.append({
-            "term": t.get("prefLabel", ""),
-            "ontology": t.get("links", {}).get("ontology", ""),
-            "mappings": mappings
-        })
-    return mapped_results
+    results = search_ontology(term)
+    cuis = []
+    for res in results:
+        if res.get("cui"):
+            cuis.extend(res["cui"])
+    return list(set(cuis))
 
 # -------------------------
-# Integration Helpers
+# Cross-Domain Semantic Context
 # -------------------------
-def ontology_to_entities(query: str) -> Dict:
+def ontology_context(term: str) -> Dict:
     """
-    Map ontology term → related drugs (ChEMBL) + genes (NCBI) + literature (PubMed).
-    This is where cross-domain linking happens.
+    Return ontology info + UMLS mapping + linked literature from PubMed.
     """
-    from genesis.api_clients import chembl_api, pubmed_api, ncbi_api  # Lazy import to avoid cycles
+    from genesis.api_clients import pubmed_api  # Lazy import
 
-    mapped_data = {
-        "ontology_term": query,
-        "drugs": [],
-        "genes": [],
-        "publications": []
-    }
-
-    # Ontology search
-    terms = search_terms(query)
-    if terms:
-        mapped_data["ontology_id"] = terms[0].get("@id", "")
-
-    # Link to ChEMBL molecules
-    mapped_data["drugs"] = chembl_api.search_molecule(query)
+    ontology_terms = search_ontology(term)
+    cuis = []
+    for res in ontology_terms:
+        if res.get("cui"):
+            cuis.extend(res["cui"])
 
-    # Link to NCBI genes
-    mapped_data["genes"] = ncbi_api.search_gene(query)
-
-    # Link to PubMed literature
-    mapped_data["publications"] = pubmed_api.search_pubmed(query)
-
-    return mapped_data
+    return {
+        "ontology_terms": ontology_terms,
+        "umls_cuis": list(set(cuis)),
+        "literature": pubmed_api.search_pubmed(term)
+    }