about summary refs log tree commit diff
path: root/gn3
diff options
context:
space:
mode:
Diffstat (limited to 'gn3')
-rw-r--r--gn3/api/lmdb_sample_data.py40
-rw-r--r--gn3/app.py2
-rw-r--r--gn3/settings.py6
3 files changed, 45 insertions, 3 deletions
diff --git a/gn3/api/lmdb_sample_data.py b/gn3/api/lmdb_sample_data.py
new file mode 100644
index 0000000..eaa71c2
--- /dev/null
+++ b/gn3/api/lmdb_sample_data.py
@@ -0,0 +1,40 @@
+"""API endpoint for retrieving sample data from LMDB storage"""
+import hashlib
+from pathlib import Path
+
+import lmdb
+from flask import Blueprint, current_app, jsonify
+
+lmdb_sample_data = Blueprint("lmdb_sample_data", __name__)
+
+
+@lmdb_sample_data.route("/sample-data/<string:dataset>/<int:trait_id>", methods=["GET"])
+def get_sample_data(dataset: str, trait_id: int):
+    """Retrieve sample data from LMDB for a given dataset and trait.
+
+    Path Parameters:
+        dataset: The name of the dataset
+        trait_id: The ID of the trait
+
+    Returns:
+        JSON object mapping sample IDs to their values
+    """
+    checksum = hashlib.md5(
+        f"{dataset}-{trait_id}".encode()
+    ).hexdigest()
+
+    db_path = Path(current_app.config["LMDB_DATA_PATH"]) / checksum
+    if not db_path.exists():
+        return jsonify(error="No data found for given dataset and trait"), 404
+    try:
+        with lmdb.open(str(db_path), max_dbs=15, readonly=True) as env:
+            data = {}
+            with env.begin(write=False) as txn:
+                cursor = txn.cursor()
+                for key, value in cursor:
+                    data[key.decode()] = float(value.decode())
+
+            return jsonify(data)
+
+    except lmdb.Error as err:
+        return jsonify(error=f"LMDB error: {str(err)}"), 500
diff --git a/gn3/app.py b/gn3/app.py
index 3841396..74bb5ab 100644
--- a/gn3/app.py
+++ b/gn3/app.py
@@ -30,6 +30,7 @@ from gn3.api.llm import gnqa
 from gn3.api.rqtl2 import rqtl2
 from gn3.api.streaming import streaming
 from gn3.case_attributes import caseattr
+from gn3.api.lmdb_sample_data import lmdb_sample_data
 
 
 class ConfigurationError(Exception):
@@ -111,6 +112,7 @@ def create_app(config: Union[Dict, str, None] = None) -> Flask:
     app.register_blueprint(gnqa, url_prefix="/api/llm")
     app.register_blueprint(rqtl2, url_prefix="/api/rqtl2")
     app.register_blueprint(streaming, url_prefix="/api/stream")
+    app.register_blueprint(lmdb_sample_data, url_prefix="/api/lmdb")
 
     register_error_handlers(app)
     return app
diff --git a/gn3/settings.py b/gn3/settings.py
index 9a3f7eb..e988106 100644
--- a/gn3/settings.py
+++ b/gn3/settings.py
@@ -19,9 +19,9 @@ SPARQL_ENDPOINT = os.environ.get(
     "SPARQL_ENDPOINT",
     "http://localhost:9082/sparql")
 
-# LMDB path
-LMDB_PATH = os.environ.get(
-    "LMDB_PATH", f"{os.environ.get('HOME')}/tmp/dataset")
+# LMDB paths
+LMDB_DATA_PATH = os.environ.get(
+    "LMDB_DATA_PATH", "/export5/lmdb-data-hashes")
 
 # SQL confs
 SQL_URI = os.environ.get(