updated pipeline.ipynb to prevent conflict

Strexas · Sep 10, 2024 · 68b8472 · 68b8472
1 parent 5073938
commit 68b8472
Showing 1 changed file with 31 additions and 3 deletions.
diff --git a/tests/pipeline.ipynb b/tests/pipeline.ipynb
@@ -238,10 +238,38 @@
   {
    "metadata": {},
    "cell_type": "code",
-   "source": "",
-   "id": "6f0abfb50bd211a0",
    "outputs": [],
-   "execution_count": null
+   "execution_count": null,
+   "source": [
+    "from api.data.refactoring import merge_gnomad_lovd, parse_gnomad, set_gnomad_dtypes\n",
+    "import pandas as pd\n",
+    "from api import (store_database_for_eys_gene,\n",
+    "                 parse_lovd,\n",
+    "                 set_lovd_dtypes,\n",
+    "                 LOVD_PATH,\n",
+    "                 GNOMAD_PATH)\n",
+    "\n",
+    "store_database_for_eys_gene('lovd', False)\n",
+    "store_database_for_eys_gene('gnomad', False)\n",
+    "\n",
+    "lovd_data = parse_lovd(LOVD_PATH + \"/lovd_data.txt\")\n",
+    "gnomad_data = parse_gnomad(GNOMAD_PATH+'/gnomad_data.csv')\n",
+    "\n",
+    "set_lovd_dtypes(lovd_data)\n",
+    "set_gnomad_dtypes(gnomad_data)\n",
+    "\n",
+    "variants_on_genome = lovd_data[\"Variants_On_Genome\"].copy()\n",
+    "\n",
+    "lovd_data = pd.merge(lovd_data[\"Variants_On_Transcripts\"],\n",
+    "                       variants_on_genome[['id','VariantOnGenome/DNA','VariantOnGenome/DNA/hg38']],\n",
+    "                       on='id',\n",
+    "                       how='left')\n",
+    "\n",
+    "gnomad_data = gnomad_data.copy()\n",
+    "final_data = merge_gnomad_lovd(lovd_data, gnomad_data)\n",
+    "final_data"
+   ],
+   "id": "ba435cd29d565f7d"
   }
  ],
  "metadata": {