Add v6.1 data and fix MSK Moorman et al pub (#706)

* Add v6.1 data and fix MSK Moorman et al pub * remove v6 files, replace v6_0 references with v6_1 --------- Co-authored-by: Onur Sumer <[email protected]>
ncihtan · Nov 5, 2024 · c1f50c6 · c1f50c6
1 parent 9dd0173
commit c1f50c6
Show file tree

Hide file tree

Showing 15 changed files with 6,029 additions and 295 deletions.
diff --git a/README.md b/README.md
@@ -20,10 +20,10 @@ that using these commands (requires access to the htan-dcc google project):
 
 ```bash
 cd data
-bq extract --destination_format CSV released.entities_v6_0 gs://htan-release-files/entities_v6_0.csv
-bq extract --destination_format CSV released.metadata_v6_0 gs://htan-release-files/metadata_v6_0.csv
-gsutil cp gs://htan-release-files/entities_v6_0.csv entities_v6_0.csv
-gsutil cp gs://htan-release-files/metadata_v6_0.csv metadata_v6_0.csv
+bq extract --destination_format CSV released.entities_v6_1 gs://htan-release-files/entities_v6_1.csv
+bq extract --destination_format CSV released.metadata_v6_1 gs://htan-release-files/metadata_v6_1.csv
+gsutil cp gs://htan-release-files/entities_v6_1.csv entities_v6_1.csv
+gsutil cp gs://htan-release-files/metadata_v6_1.csv metadata_v6_1.csv
 
 ```
 

diff --git a/components/HomePage.tsx b/components/HomePage.tsx
@@ -54,7 +54,7 @@ const HomePage: React.FunctionComponent<IHomePropsProps> = ({
                     }}
                 >
                     <a style={{ color: 'white' }} href="/data-updates">
-                        Data Release V6.0 (Last updated 2024-08-26)
+                        Data Release V6.1 (Last updated 2024-11-05)
                     </a>
                 </div>
                 <Row className="justify-content-md-center">

diff --git a/data/entities_v6_0.csv → data/entities_v6_1.csv b/data/entities_v6_0.csv → data/entities_v6_1.csv
diff --git a/data/get_syn_data.py b/data/get_syn_data.py
@@ -103,13 +103,13 @@ def generate_json(include_at_risk_populations, include_released_only, do_not_dow
     }
 
     if include_released_only:
-        released_entities_df = pd.read_csv("entities_v6_0.csv")
+        released_entities_df = pd.read_csv("entities_v6_1.csv")
         include_release_ids = set(released_entities_df['entityId'])
 
     # store all metadata synapse ids for downloading submitted metadata directly
     portal_metadata = {}
 
-    released_metadata_df = pd.read_csv("metadata_v6_0.csv")
+    released_metadata_df = pd.read_csv("metadata_v6_1.csv")
     released_synapse_metadata_ids = set(released_metadata_df['Manifest_Id'])
 
     # iterate over projects; map to HTAN ID, inspect metadata and add to portal JSON dump
@@ -339,4 +339,4 @@ def generate_json(include_at_risk_populations, include_released_only, do_not_dow
 
 
 if __name__ == '__main__':
-    generate_json()
+    generate_json()