EVA-3304 - Validate the presence of the collection location and date (#…

…155) * Validate the presence of the collection location and date * Update test Co-authored-by: April Shen <april.tuesday@gmail.com>
EBIvariation · Jun 14, 2023 · 7c9fa0c · 7c9fa0c
1 parent 673280c
commit 7c9fa0c
Show file tree

Hide file tree

Showing 7 changed files with 17 additions and 3 deletions.
diff --git a/eva_submission/biosamples_submission.py b/eva_submission/biosamples_submission.py
@@ -348,6 +348,11 @@ class SampleMetadataSubmitter(SampleSubmitter):
                                 'ecotype', 'isolate', 'strain', 'sub_species', 'variety', 'sub_strain', 'cell_line',
                                 'serotype', 'serovar']
 
+    characteristic_defaults = {
+        'collection_date': 'not provided',
+        'geographic location (country and/or sea)': 'not provided'
+    }
+
     submitter_mapping = {
         'Email Address': 'E-mail',
         'First Name': 'FirstName',
@@ -405,7 +410,14 @@ def map_metadata_to_bsd_data(self):
                         self.map_sample_key(attribute.lower()),
                         [{'text': self.serialize(value)}]
                     )
-
+            # Apply defaults if the key doesn't already exist
+            for key in self.characteristic_defaults:
+                if key not in sample_row:
+                    self.apply_mapping(
+                        bsd_sample_entry['characteristics'],
+                        self.map_sample_key(key.lower()),
+                        [{'text': self.serialize(self.characteristic_defaults[key])}]
+                    )
             project_row = self.reader.project
             for key in self.reader.project:
                 if key in self.project_mapping:

diff --git a/eva_submission/xlsx/xlsx_validation.py b/eva_submission/xlsx/xlsx_validation.py
@@ -72,7 +72,8 @@ def complex_validation(self):
             self.group_of_fields_required(
                 'Sample', row,
                 ['Analysis Alias', 'Sample Accession', 'Sample ID'],
-                ['Analysis Alias', 'Sample Name', 'Title', 'Tax Id', 'Scientific Name']
+                ['Analysis Alias', 'Sample Name', 'Title', 'Tax Id', 'Scientific Name', 'collection_date',
+                 'geographic location (country and/or sea)']
             )
 
     def semantic_validation(self):

diff --git a/tests/resources/brokering/metadata_sheet2.xlsx b/tests/resources/brokering/metadata_sheet2.xlsx
diff --git a/tests/resources/brokering/metadata_sheet_fail.xlsx b/tests/resources/brokering/metadata_sheet_fail.xlsx
diff --git a/tests/resources/brokering/metadata_wrong_scientific_name.xlsx b/tests/resources/brokering/metadata_wrong_scientific_name.xlsx
diff --git a/tests/resources/metadata.xlsx b/tests/resources/metadata.xlsx
diff --git a/tests/test_submit_to_bsd.py b/tests/test_submit_to_bsd.py
@@ -294,7 +294,8 @@ def test_map_metadata_to_bsd_data(self):
                 'Organism': [{'text': 'Homo sapiens'}],
                 'description': [{'text': 'Sample %s' % (i+1)}],
                 'scientific name': [{'text': 'Homo sapiens'}],
-                'collection_date': [{'text': '2020-01-15'}]
+                'collection_date': [{'text': '2020-01-15'}],
+                'geographic location (country and/or sea)': [{'text': 'not provided'}]
             }}
             for i in range(100)
         ]