VLE/KBE-34/routing_of_merging_data

api/data/__init__.py

@@ -33,26 +33,6 @@
   GNOMAD_PATH,
 )
 
-# DATA COLLECTION IMPORT
-from .collection import (
-    # Custom exceptions
-    BadResponseException,
-    DownloadError,
-
-    # Custom utility functions
-    get_file_from_url,
-
-    # Functions for downloading databases
-    download_lovd_database_for_eys_gene,
-    download_genes_lovd,
-    download_database_for_eys_gene,
-    download_data_from_gnomad_eys,
-
-    # Functions for storing databases
-    store_database_for_eys_gene
-
-)
-
 # DATA REFACTORING IMPORT
 from .refactoring import (
     # Functions for refactoring data

api/data/constants.py

@@ -25,6 +25,7 @@
 LOVD_PATH = os.path.join(DATA_PATH, "lovd/")
 GNOMAD_PATH = os.path.join(DATA_PATH, "gnomad/")
 CLINVAR_PATH = os.path.join(DATA_PATH, "clinvar/")
+DEFAULT_SAVE_PATH = os.path.join(DATA_PATH, "merged_data/")
 
 # variable data types
 LOVD_TABLES_DATA_TYPES = {

api/data/refactoring.py

@@ -4,15 +4,14 @@
 import logging
 import re
 
-import requests
 
 import pandas as pd
 from pandas import DataFrame
 
 from pyliftover import LiftOver
 
-from .constants import LOVD_TABLES_DATA_TYPES, LOVD_PATH, GNOMAD_TABLES_DATA_TYPES, GNOMAD_PATH
-
+from .constants import LOVD_TABLES_DATA_TYPES, LOVD_PATH, GNOMAD_TABLES_DATA_TYPES, GNOMAD_PATH, \
+    DEFAULT_SAVE_PATH
 
 
 def set_lovd_dtypes(df_dict):
@@ -347,4 +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]
-

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__)
 
 
@@ -85,8 +90,41 @@ def get_workspace_merge_lovd_gnomad(relative_path):
         # [destination_path, override, lovd_file, gnomad_file]
         #
 
-        # TODO: Remove this sleep statement once the merge logic is implemented
-        time.sleep(1)  # Simulate a delay for the merge process
+        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}")
 
         # Emit a feedback to the user's console
         socketio_emit_to_user_session(