Start adding provenance - Add github URL with specified tag. (#353)

* Add in tag or commit to GENIE storing of files

* update tag

* use version stated in GENIE package as tag for provenance

* make sure data guide has provenance added

bin/database_to_staging.py

@@ -284,9 +284,9 @@ def main(genie_version,
                                          database_mapping=databaseSynIdMappingId,
                                          genie_user=genie_user,
                                          genie_pass=genie_pass)
-    data_guide_ent = synapseclient.File(data_guide_pdf,
-                                        parent=folders['release_folder'])
-    syn.store(data_guide_ent)
+    database_to_staging.store_file(syn, data_guide_pdf,
+                                   genieVersion=genie_version,
+                                   parent=folders['release_folder'])
     logger.info("COMPLETED DATABASE TO STAGING")
 
 

genie/database_to_staging.py

@@ -14,6 +14,7 @@
 import synapseutils
 
 from . import process_functions
+from . import __version__
 
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
@@ -61,7 +62,8 @@ def find_caselistid(syn, parentid):
 
 
 def store_file(syn, filePath, genieVersion="database", name=None,
-               parent=None, fileFormat=None, cBioFileFormat=None):
+               parent=None, fileFormat=None, cBioFileFormat=None,
+               tag_or_commit=None):
     '''
     Convenience function to store files
 
@@ -74,6 +76,7 @@ def store_file(syn, filePath, genieVersion="database", name=None,
         cBioFileFormat: cBioPortal file format
         staging: Staging GENIE release.  Default to False
         caseLists: Case lists are stored elsewhere
+        tag_or_commit: Github tag or commit
     '''
     logger.info("STORING FILE: {}".format(os.path.basename(filePath)))
     if name is None:
@@ -84,7 +87,12 @@ def store_file(syn, filePath, genieVersion="database", name=None,
         ent.fileFormat = fileFormat
     if cBioFileFormat is not None:
         ent.cBioFileFormat = cBioFileFormat
-    ent = syn.store(ent)
+    if tag_or_commit is None:
+        tag_or_commit = f"v{__version__.__version__}"
+    ent = syn.store(
+        ent,
+        used=f"https://github.com/Sage-Bionetworks/Genie/tree/{tag_or_commit}"
+    )
     return ent