move content of routing merge to app/back-end/src/routes/workspace_me…

…rge_route
Strexas · Sep 25, 2024 · 1b6ac7d · 1b6ac7d
1 parent 2402ea6
commit 1b6ac7d
Show file tree

Hide file tree

Showing 3 changed files with 41 additions and 95 deletions.
diff --git a/api/data/refactoring.py b/api/data/refactoring.py
@@ -346,56 +346,3 @@ def find_popmax_in_gnomad(data):
                 max_id = population_id
         data.loc[i, 'Popmax'] = max_pop
         data.loc[i, 'Popmax population'] = population_mapping[max_id]
-
-
-def routing_merge(lovd_path:str=LOVD_PATH,
-                  gnomad_path:str=GNOMAD_PATH,
-                  save_path:str=DEFAULT_SAVE_PATH,
-                  overwrite:bool=False):
-    """
-    Merges data from provided paths and saves to new location
-    :param overwrite: does file requires overwriting
-    :param lovd_path: path to LOVD dataframe
-    :param gnomad_path: path to gnomAD dataframe
-    :param save_path: path where to save merged data
-    :return: None
-    """
-
-    save_as = os.path.join(save_path, "lovd_gnomad.csv")
-
-    if os.path.exists(save_as) and not overwrite:
-        return
-
-    if not os.path.exists(save_path):
-        os.makedirs(save_path)
-
-    if not os.path.exists(os.path.join(lovd_path, "lovd_data.txt")):
-        raise FileNotFoundError(f"LOVD data file not found at: {lovd_path}")
-
-    if not os.path.exists(os.path.join(gnomad_path, "gnomad_data.csv")):
-        raise FileNotFoundError(f"gnomAD data file not found at: {gnomad_path}")
-
-    lovd_data = parse_lovd(lovd_path + "/lovd_data.txt")
-    gnomad_data = parse_gnomad(gnomad_path + '/gnomad_data.csv')
-
-    set_lovd_dtypes(lovd_data)
-    set_gnomad_dtypes(gnomad_data)
-
-    # Extract "Variants_On_Genome" and merge it with "Variants_On_Transcripts"
-    variants_on_genome = lovd_data["Variants_On_Genome"].copy()
-    gnomad_data = gnomad_data.copy()
-
-    lovd_data = pd.merge(
-        lovd_data["Variants_On_Transcripts"],
-        variants_on_genome[['id', 'VariantOnGenome/DNA', 'VariantOnGenome/DNA/hg38']],
-        on='id',
-        how='left'
-    )
-
-    final_data = merge_gnomad_lovd(lovd_data, gnomad_data)
-
-    try:
-        final_data.to_csv(save_as)
-        print(f"Merged data saved to {save_path}")
-    except OSError as e:
-        print(f"Error saving file: {e}")
diff --git a/app/back-end/src/routes/workspace_merge_route.py b/app/back-end/src/routes/workspace_merge_route.py
@@ -8,6 +8,8 @@
 
 import os
 import time  # TODO: Remove this import once the merge logic is implemented
+
+import pandas as pd
 from flask import Blueprint, request, jsonify
 
 from src.setup.extensions import logger
@@ -20,6 +22,9 @@
     WORKSPACE_UPDATE_FEEDBACK_EVENT,
 )
 
+from api import set_lovd_dtypes, parse_lovd
+from api.data import merge_gnomad_lovd, set_gnomad_dtypes, parse_gnomad
+
 workspace_merge_route_bp = Blueprint("workspace_merge_route", __name__)
 
 
@@ -85,6 +90,42 @@ def get_workspace_merge_lovd_gnomad(relative_path):
         # [destination_path, override, lovd_file, gnomad_file]
         #
 
+        if os.path.exists(destination_path) and not override:
+            return
+
+        if not os.path.exists(destination_path):
+            os.makedirs(destination_path)
+
+        if not os.path.exists(lovd_file):
+            raise FileNotFoundError(f"LOVD data file not found at: {lovd_file}")
+
+        if not os.path.exists(gnomad_file):
+            raise FileNotFoundError(f"gnomAD data file not found at: {gnomad_file}")
+
+        lovd_data = parse_lovd(lovd_file)
+        gnomad_data = parse_gnomad(gnomad_file)
+
+        set_lovd_dtypes(lovd_data)
+        set_gnomad_dtypes(gnomad_data)
+
+        # Extract "Variants_On_Genome" and merge it with "Variants_On_Transcripts"
+        variants_on_genome = lovd_data["Variants_On_Genome"].copy()
+        gnomad_data = gnomad_data.copy()
+
+        lovd_data = pd.merge(
+            lovd_data["Variants_On_Transcripts"],
+            variants_on_genome[['id', 'VariantOnGenome/DNA', 'VariantOnGenome/DNA/hg38']],
+            on='id',
+            how='left'
+        )
+
+        final_data = merge_gnomad_lovd(lovd_data, gnomad_data)
+
+        try:
+            final_data.to_csv(destination_path)
+        except OSError as e:
+            raise RuntimeError(f"Error saving file: {e}")
+
         # TODO: Remove this sleep statement once the merge logic is implemented
         time.sleep(1)  # Simulate a delay for the merge process
 

diff --git a/tests/pipeline.ipynb b/tests/pipeline.ipynb
@@ -2181,48 +2181,6 @@
    ],
    "id": "d86fa6b925aea085",
    "execution_count": null
-  },
-  {
-   "metadata": {
-    "ExecuteTime": {
-     "end_time": "2024-09-22T13:05:20.884141Z",
-     "start_time": "2024-09-22T13:05:19.105877Z"
-    }
-   },
-   "cell_type": "code",
-   "source": [
-    "from api.data.refactoring import routing_merge\n",
-    "routing_merge()"
-   ],
-   "id": "29ecf5e58e3d53e4",
-   "outputs": [],
-   "execution_count": 1
-  },
-  {
-   "metadata": {
-    "ExecuteTime": {
-     "end_time": "2024-09-22T13:03:46.634257Z",
-     "start_time": "2024-09-22T13:02:58.726380Z"
-    }
-   },
-   "cell_type": "code",
-   "source": [
-    "from api import store_database_for_eys_gene\n",
-    "\n",
-    "store_database_for_eys_gene('lovd', False)\n",
-    "store_database_for_eys_gene('gnomad', False)"
-   ],
-   "id": "b5eedffd56faee1d",
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "The file at ../data/lovd/lovd_data.txt already exists.\n"
-     ]
-    }
-   ],
-   "execution_count": 1
   }
  ],
  "metadata": {