Merge pull request #108 from 4dn-dcic/fix_geo_submission_script

Fix geo submission script

CHANGELOG.rst

@@ -6,6 +6,11 @@ dcicwrangling
 Change Log
 ----------
 
+2.4.1
+=====
+
+* Fixed 12_geo_submission.ipynb to avoid adding duplicate protocol items in the JSON files specifically for experiments where both 'protocol' and 'protocol_variation' is present
+
 2.4.0
 =====
 

notebooks/useful_notebooks/12_geo_submission.ipynb

@@ -256,9 +256,12 @@
     "    if experiment_object.get('protocol'):\n",
     "        protocol = get_item_in_store(experiment_object['protocol'])\n",
     "        protocols.append(protocol)\n",
-    "    if experiment_object.get('protocol_variations'):\n",
-    "        protocol_variations = [p for p in get_item_in_store(experiment_object['protocol_variations'])]\n",
-    "        protocols.extend(protocol_variations)\n",
+    "    if experiment_object['protocol_variation']:\n",
+    "        protocol_variations = experiment_object['protocol_variation']\n",
+    "        for p in protocol_variations:\n",
+    "            protocol_id = p['@id']\n",
+    "            protocol_variations_details = get_item_in_store(protocol_id)\n",
+    "            protocols.append(protocol_variations_details)\n",
     "    if experiment_object.get('cell_sorting_protocol'):\n",
     "        protocol_sorting = get_item_in_store(experiment_object['cell_sorting_protocol'])\n",
     "        protocols.append(protocol_sorting)\n",
@@ -378,12 +381,17 @@
     "    experiment_dict = {}\n",
     "    file_ids = []\n",
     "    biosample_id = ''\n",
+    "    all_keys = []\n",
     "    for key, value in experiment_object.items():\n",
     "        export_value = None\n",
+    "        all_keys.append(key)\n",
     "        if key in experiment_simple_values:\n",
     "            export_value = value\n",
-    "        elif key in ['biosample_quantity', 'protocol', 'protocol_variation', 'cell_sorting_protocol']:\n",
+    "        elif key in ['biosample_quantity', 'protocol', 'cell_sorting_protocol']:\n",
     "            export_value = experiment_function_dispatch[key](experiment_object)\n",
+    "        elif key == 'protocol_variation':\n",
+    "            if 'protocol' not in all_keys:\n",
+    "                export_value = experiment_function_dispatch[key](experiment_object)\n",
     "        elif key in ['files', 'processed_files', 'other_processed_files', 'reference_files']:\n",
     "            export_value, files_list = experiment_function_dispatch[key](value)\n",
     "            file_ids.extend(files_list)\n",
@@ -670,7 +678,12 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.7.2"
+   "version": "3.8.12"
+  },
+  "vscode": {
+   "interpreter": {
+    "hash": "20d91ac981d81ffaf62d1b59390659afd05fbcd41a0bec0e13249d20dc131a1e"
+   }
   }
  },
  "nbformat": 4,

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "dcicwrangling"
-version = "2.4.0"
+version = "2.4.1"
 description = "Scripts and Jupyter notebooks for 4DN wrangling"
 authors = ["4DN-DCIC Team <[email protected]>"]
 license = "MIT"