Update genesis/api_clients/ncbi_api.py

genesis/api_clients/ncbi_api.py

@@ -137,17 +137,53 @@ def search_proteins(term: str, retmax: int = 5) -> List[Dict]:
     proteins = [{"id": pid, "fasta": fasta_data} for pid in ids]
     return proteins
 
+# -------------------------
+# Structure Search
+# -------------------------
+def fetch_ncbi_structure(term: str, retmax: int = 5) -> List[Dict]:
+    """
+    Search NCBI Structure database and return structure metadata.
+    """
+    ids = ncbi_search("structure", term, retmax)
+    if not ids:
+        return []
+
+    params = {
+        "db": "structure",
+        "id": ",".join(ids),
+        "retmode": "json"
+    }
+    if NCBI_API_KEY:
+        params["api_key"] = NCBI_API_KEY
+
+    r = session.get(f"{NCBI_BASE}/esummary.fcgi", params=params)
+    r.raise_for_status()
+
+    records = r.json().get("result", {})
+    structures = []
+    for sid in ids:
+        rec = records.get(sid, {})
+        structures.append({
+            "structure_id": sid,
+            "title": rec.get("title"),
+            "organism": rec.get("organism"),
+            "release_date": rec.get("releasedate"),
+            "link": f"https://www.ncbi.nlm.nih.gov/structure/{sid}"
+        })
+    return structures
+
 # -------------------------
 # Build Cross-Database Profile
 # -------------------------
 def ncbi_cross_profile(term: str) -> Dict:
     """
-    Given a term, pull literature, genes, and proteins for unified output.
+    Given a term, pull literature, genes, proteins, and structures for unified output.
     """
     return {
         "term": term,
         "timestamp": datetime.utcnow().isoformat(),
         "literature": search_pubmed(term, retmax=5),
         "genes": search_genes(term, retmax=5),
-        "proteins": search_proteins(term, retmax=2)
+        "proteins": search_proteins(term, retmax=2),
+        "structures": fetch_ncbi_structure(term, retmax=3)
     }