add relatedness check tsv to failure message (#4456)

* add relatedness check tsv to failure message * Update check_for_new_samples_from_pipeline_tests.py
broadinstitute · Nov 6, 2024 · d84fef0 · d84fef0
1 parent bea7763
commit d84fef0
Show file tree

Hide file tree

Showing 2 changed files with 12 additions and 3 deletions.
diff --git a/seqr/management/commands/check_for_new_samples_from_pipeline.py b/seqr/management/commands/check_for_new_samples_from_pipeline.py
@@ -29,6 +29,7 @@
 DATASET_TYPE_MAP = {'GCNV': Sample.DATASET_TYPE_SV_CALLS}
 USER_EMAIL = 'manage_command'
 MAX_LOOKUP_VARIANTS = 5000
+RELATEDNESS_CHECK_NAME = 'relatedness_check'
 
 PDO_COPY_FIELDS = [
     'PDO', 'PDOStatus', 'SeqrLoadingDate', 'GATKShortReadCallsetPath', 'SeqrProjectURL', 'TerraProjectURL',
@@ -160,6 +161,7 @@ def _load_new_samples(cls, metadata_path, genome_version, dataset_type, run_vers
                 split_project_pdos[project.name] = cls._update_pdos(session, project.guid, sample_ids)
 
         # Send failure notifications
+        relatedness_check_file_path = metadata.get('relatedness_check_file_path')
         failed_family_samples = metadata.get('failed_family_samples', {})
         failed_families_by_guid = {f['guid']: f for f in Family.objects.filter(
             guid__in={family for families in failed_family_samples.values() for family in families}
@@ -176,9 +178,15 @@ def _load_new_samples(cls, metadata_path, genome_version, dataset_type, run_vers
                 split_pdos = split_project_pdos.get(project)
                 if split_pdos:
                     summary += f'\n\nSkipped samples in this project have been moved to {", ".join(split_pdos)}'
+
+                relatedness_check_message = (
+                    f'\nRelatedness check results: {relatedness_check_file_path}'
+                    if (relatedness_check_file_path and check == RELATEDNESS_CHECK_NAME)
+                    else ''
+                )
                 safe_post_to_slack(
                     SEQR_SLACK_LOADING_NOTIFICATION_CHANNEL,
-                    f'The following {len(failures)} families failed {check.replace("_", " ")} in {project}:\n{summary}'
+                    f'The following {len(failures)} families failed {check.replace("_", " ")} in {project}:\n{summary}{relatedness_check_message}'
                 )
 
         # Reload saved variant JSON

diff --git a/seqr/management/tests/check_for_new_samples_from_pipeline_tests.py b/seqr/management/tests/check_for_new_samples_from_pipeline_tests.py
@@ -169,7 +169,8 @@
             'F000002_2': {'reasons': ["Missing samples: {'HG00732', 'HG00733'}"]},
             'F000003_3': {'reasons': ["Missing samples: {'NA20870'}"]},
         },
-    }
+    },
+    'relatedness_check_file_path': 'gs://seqr-loading-temp/v3.1/GRCh38/SNV_INDEL/relatedness_check/test_callset_hash.tsv',
 }, {
     'callsets': ['invalid_family.vcf'],
     'sample_type': 'WGS',
@@ -552,7 +553,7 @@ def test_command(self, mock_email, mock_airtable_utils):
             mock.call(
                 'seqr_loading_notifications',
                 """The following 1 families failed relatedness check in 1kg project nåme with uniçøde:
-- 1: Sample NA19679 has expected relation "parent" to NA19675 but has coefficients [0.0, 0.8505002045292791, 0.14949979547072176, 0.5747498977353613]; Sample NA19678 has expected relation "sibling" to NA19675 but has coefficients [0.17424888135104177, 0.6041745754450025, 0.22157654320395614, 0.5236638309264574]""",
+- 1: Sample NA19679 has expected relation "parent" to NA19675 but has coefficients [0.0, 0.8505002045292791, 0.14949979547072176, 0.5747498977353613]; Sample NA19678 has expected relation "sibling" to NA19675 but has coefficients [0.17424888135104177, 0.6041745754450025, 0.22157654320395614, 0.5236638309264574]\nRelatedness check results: gs://seqr-loading-temp/v3.1/GRCh38/SNV_INDEL/relatedness_check/test_callset_hash.tsv""",
             ),
             mock.call(
                 'seqr_loading_notifications',