Fix from submissions (#166)

* Reformat the archival notification text
* Make the brokering results overwrite previous results for everything but PROJECT

eva_submission/eload_brokering.py

@@ -84,10 +84,12 @@ def broker_to_ena(self, force=False, existing_project=None, async_upload=False,
             # Upload XML to ENA
             ena_uploader.upload_xml_files_to_ena(dry_ena_upload)
             if not dry_ena_upload:
-                # Update the accessions in case we're working with existing project
-                accessions = ena_uploader.results
-                accessions.update(self.eload_cfg.query('brokering', 'ena', ret_default={}))
-                self.eload_cfg.set('brokering', 'ena', value=accessions)
+                # Update the project accession in case we're working with existing project
+                # We should not be uploading additional analysis in th same ELOAD so no need to update
+                pre_existing_project = self.eload_cfg.query('brokering', 'ena', 'PROJECT')
+                if pre_existing_project and 'PROJECT' not in ena_uploader.results:
+                    ena_uploader.results['PROJECT'] = pre_existing_project
+                self.eload_cfg.set('brokering', 'ena', value=ena_uploader.results)
                 self.eload_cfg.set('brokering', 'ena', 'date', value=self.now)
                 self.eload_cfg.set('brokering', 'ena', 'hold_date', value=ena_uploader.converter.hold_date)
                 self.eload_cfg.set('brokering', 'ena', 'pass', value=not bool(ena_uploader.results['errors']))
@@ -292,8 +294,9 @@ def _ena_report(self):
         return '\n'.join(reports)
 
     def _archival_confirmation_text(self):
+        if not self._brokering_complete():
+            return 'NA'
         study_title = self.eload_cfg.query('submission', 'project_title')
-
         hold_date = self.eload_cfg.query('brokering', 'ena', 'hold_date')
         brokering_date_from_config = self.eload_cfg.query('brokering', 'brokering_date')
         try:
@@ -302,18 +305,24 @@ def _archival_confirmation_text(self):
             brokering_date = datetime.date.today()
         brokering_date_plus_3 = brokering_date + datetime.timedelta(days=3)
         available_date = hold_date if hold_date is not None else brokering_date_plus_3
-
+        if isinstance(available_date, datetime.datetime) or isinstance(available_date, datetime.date):
+            available_date_str = available_date.strftime("%Y-%m-%d")
+        else:
+            available_date_str = available_date.split(" ")[0]
         project_accession = self.eload_cfg.query('brokering', 'ena', 'PROJECT')
-        analysis_accession = self.eload_cfg.query('brokering', 'ena', 'ANALYSIS')
+        analysis_accession = self.eload_cfg.query('brokering', 'ena', 'ANALYSIS', ret_default={})
 
         taxonomy_id = self.eload_cfg.query('submission', 'taxonomy_id')
         non_human_study_text = 'Please allow at least 48 hours from the initial release date provided for the data to be made available through this link. Each variant will be issued a unique SS# ID which will be made available to download via the "browsable files" link on the EVA study page.' if taxonomy_id!=9606 else ""
 
         archival_text_data = {
             'study_title': study_title,
-            'available_date': available_date,
+            'available_date': available_date_str,
             'project_accession': project_accession,
-            'analysis_accession': analysis_accession,
+            'analysis_accession': ', '.join([
+                f'{self._undo_unique_alias(alias)}=>{accession}'
+                for alias, accession in analysis_accession.items()
+            ]),
             'non_human_study': non_human_study_text
         }
 
@@ -326,7 +335,7 @@ def _archival_confirmation_text(self):
 You can also notify us when your paper has been assigned a PMID. We will add this to your study page in the EVA. If there is anything else you need please do not hesitate to notify me. Archived data can be referenced using the project accession & associated URL e.g. The variant data for this study have been deposited in the European Variation Archive (EVA) at EMBL-EBI under accession number {project_accession} (https://www.ebi.ac.uk/eva/?eva-study={project_accession})
 The EVA can be cited directly using the associated literature:
 Cezard T, Cunningham F, Hunt SE, Koylass B, Kumar N, Saunders G, Shen A, Silva AF, Tsukanov K, Venkataraman S, Flicek P, Parkinson H, Keane TM. The European Variation Archive: a FAIR resource of genomic variation for all species. Nucleic Acids Res. 2021 Oct 28:gkab960. doi: 10.1093/nar/gkab960. PMID: 34718739.
-        """
+"""
 
         return archival_text.format(**archival_text_data)
 
@@ -357,3 +366,6 @@ def report(self):
 {archival_confirmation_text}
 """
         print(report.format(**report_data))
+
+    def _brokering_complete(self):
+        return all([self.eload_cfg.query('brokering', key, 'pass') for key in ['ena', 'Biosamples']])

eva_submission/eload_submission.py

@@ -79,6 +79,11 @@ def _unique_alias(self, alias):
             return f'{self.eload}_{alias}'
         return alias
 
+    def _undo_unique_alias(self, alias):
+        if alias.startswith(self.eload + '_'):
+            return alias[len(self.eload) + 1:]
+        return alias
+
     def create_log_file(self):
         logfile_name = os.path.join(self.eload_dir, str(self.eload) + "_submission.log")
         if logfile_name not in eload_logging_files:

tests/resources/eloads/ELOAD_4/.ELOAD_4_config.yml

@@ -7,7 +7,8 @@ brokering:
     pass: true
   brokering_date: 2021-01-01 12:20:.0
   ena:
-    ANALYSIS: ERZ0000001
+    ANALYSIS:
+      alias1: ERZ0000001
     PROJECT: PRJEB00001
     SUBMISSION: ERA0000001
     errors: []

tests/test_eload_brokering.py

@@ -179,7 +179,7 @@ def test_report(self):
     - Hold date: 
     - Accessions: PROJECT: PRJEB00001
 SUBMISSION: ERA0000001
-ANALYSIS: ERZ0000001
+ANALYSIS: {'alias1': 'ERZ0000001'}
     - Errors: 
     - receipt: <?xml version="1.0" encoding="UTF-8"?>
 <?xml-stylesheet type="text/xsl" href="receipt.xsl"?>
@@ -202,13 +202,13 @@ def test_report(self):
 
 Your EVA submission "Greatest project ever" has now been archived and will be made available to the public on 2021-01-04. The accessions associated with your submission are:
 Project: PRJEB00001
-Analyses: ERZ0000001 
+Analyses: alias1=>ERZ0000001 
 If you wish your data to be held private beyond the date specified above, please let us know. Once released, the data will be made available to download from this link: https://www.ebi.ac.uk/eva/?eva-study=PRJEB00001
 Please allow at least 48 hours from the initial release date provided for the data to be made available through this link. Each variant will be issued a unique SS# ID which will be made available to download via the "browsable files" link on the EVA study page.
 You can also notify us when your paper has been assigned a PMID. We will add this to your study page in the EVA. If there is anything else you need please do not hesitate to notify me. Archived data can be referenced using the project accession & associated URL e.g. The variant data for this study have been deposited in the European Variation Archive (EVA) at EMBL-EBI under accession number PRJEB00001 (https://www.ebi.ac.uk/eva/?eva-study=PRJEB00001)
 The EVA can be cited directly using the associated literature:
 Cezard T, Cunningham F, Hunt SE, Koylass B, Kumar N, Saunders G, Shen A, Silva AF, Tsukanov K, Venkataraman S, Flicek P, Parkinson H, Keane TM. The European Variation Archive: a FAIR resource of genomic variation for all species. Nucleic Acids Res. 2021 Oct 28:gkab960. doi: 10.1093/nar/gkab960. PMID: 34718739.
-        
+
 '''
         with patch('builtins.print') as mprint:
             self.existing_eload.eload_cfg.set('submission', 'metadata_spreadsheet', value=os.path.join(self.existing_eload.eload_dir, '10_submitted/metadata_file/metadata_sheet.xlsx'))
@@ -257,3 +257,22 @@ def test_update_metadata_from_config_for_files(self):
             {'Analysis Alias': 'GAE2', 'File Name': 'ELOAD_3/GAE2.vcf.gz', 'File Type': 'vcf', 'MD5': None, 'row_num': 4},
             {'Analysis Alias': 'GAE2', 'File Name': 'ELOAD_3/GAE2.vcf.gz.csi', 'File Type': 'csi', 'MD5': None, 'row_num': 5}
         ]
+
+    def test_archival_confirmation_text(self):
+        self.eload.eload_cfg.set('submission', 'project_title', value='Great project')
+        self.eload.eload_cfg.set('brokering', 'Biosamples', 'pass', value=True)
+        self.eload.eload_cfg.set('brokering', 'ena', 'pass', value=True)
+        self.eload.eload_cfg.set('brokering', 'ena', 'hold_date', value='2022-10-21 09:14:53.389160')
+        self.eload.eload_cfg.set('brokering', 'ena', 'PROJECT', value='PRJEB00001')
+        self.eload.eload_cfg.set('brokering', 'ena', 'ANALYSIS', value={'ELOAD_3_alias1': 'ERZ00000001', 'ELOAD_3_alias2': 'ERZ00000002'})
+        expected_text = '''
+Your EVA submission "Great project" has now been archived and will be made available to the public on 2022-10-21. The accessions associated with your submission are:
+Project: PRJEB00001
+Analyses: alias1=>ERZ00000001, alias2=>ERZ00000002 
+If you wish your data to be held private beyond the date specified above, please let us know. Once released, the data will be made available to download from this link: https://www.ebi.ac.uk/eva/?eva-study=PRJEB00001
+Please allow at least 48 hours from the initial release date provided for the data to be made available through this link. Each variant will be issued a unique SS# ID which will be made available to download via the "browsable files" link on the EVA study page.
+You can also notify us when your paper has been assigned a PMID. We will add this to your study page in the EVA. If there is anything else you need please do not hesitate to notify me. Archived data can be referenced using the project accession & associated URL e.g. The variant data for this study have been deposited in the European Variation Archive (EVA) at EMBL-EBI under accession number PRJEB00001 (https://www.ebi.ac.uk/eva/?eva-study=PRJEB00001)
+The EVA can be cited directly using the associated literature:
+Cezard T, Cunningham F, Hunt SE, Koylass B, Kumar N, Saunders G, Shen A, Silva AF, Tsukanov K, Venkataraman S, Flicek P, Parkinson H, Keane TM. The European Variation Archive: a FAIR resource of genomic variation for all species. Nucleic Acids Res. 2021 Oct 28:gkab960. doi: 10.1093/nar/gkab960. PMID: 34718739.
+'''
+        assert self.eload._archival_confirmation_text() == expected_text

tests/test_xlsx_validation.py

@@ -25,7 +25,6 @@ def setUp(self) -> None:
     def tearDown(self) -> None:
         os.remove(self.metadata_file_wrong_sc_name_copy)
 
-
     def test_cerberus_validation(self):
         self.validator.cerberus_validation()
         self.assertEqual(self.validator.error_list, [])
@@ -44,7 +43,7 @@ def test_complex_validation_failure(self):
         expected_errors = [
             'Check Analysis Alias vs Samples: GAE2,None present in Analysis Alias not in Samples',
             'Check Analysis Alias vs Files: GAE2,None present in Analysis Alias not in Files',
-            'In row 102, collection_date is not a date or "not provided": it is "Dote of collection"'
+            'In row 102, collection_date is not a date or "not provided": it is set to "Date of collection"'
         ]
         self.assertEqual(self.validator_fail.error_list, expected_errors)