Get_cram_paths external IDs fix (#828)

* Fix analysis table get_cram_paths function with external_id changes

* Bump version: 7.1.0 → 7.1.1

* Also rename p.id; return only the primary external_id

Add a test case that exercises this code.

* Rename table alias to the conventional 'peid.'

---------

Co-authored-by: John Marshall <[email protected]>

Author: EddieLF

.bumpversion.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 7.1.0
+current_version = 7.1.1
 commit = True
 tag = False
 parse = (?P<major>\d+)\.(?P<minor>\d+)\.(?P<patch>[A-z0-9-]+)

api/server.py

@@ -20,7 +20,7 @@
 from db.python.utils import get_logger
 
 # This tag is automatically updated by bump2version
-_VERSION = '7.1.0'
+_VERSION = '7.1.1'
 
 
 logger = get_logger()

db/python/tables/analysis.py

@@ -13,6 +13,7 @@
     to_db_json,
 )
 from models.enums import AnalysisStatus
+from models.models import PRIMARY_EXTERNAL_ORG
 from models.models.analysis import AnalysisInternal
 from models.models.audit_log import AuditLogInternal
 from models.models.project import ProjectId
@@ -413,12 +414,13 @@ async def get_sample_cram_path_map_for_seqr(
     ) -> List[dict[str, str]]:
         """Get (ext_sample_id, cram_path, internal_id) map"""
 
-        values: dict[str, Any] = {'project': project}
+        values: dict[str, Any] = {'project': project, 'PRIMARY_EXTERNAL_ORG': PRIMARY_EXTERNAL_ORG}
         filters = [
             'a.active',
             'a.type = "cram"',
             'a.status = "completed"',
-            'p.project = :project',
+            'peid.project = :project',
+            'peid.name = :PRIMARY_EXTERNAL_ORG',
         ]
         if sequencing_types:
             if len(sequencing_types) == 1:
@@ -431,16 +433,16 @@ async def get_sample_cram_path_map_for_seqr(
             filters.append('JSON_VALUE(a.meta, "$.sequencing_type") ' + seq_check)
 
         if participant_ids:
-            filters.append('p.id IN :pids')
+            filters.append('peid.participant_id IN :pids')
             values['pids'] = list(participant_ids)
 
         _query = f"""
-SELECT p.external_id as participant_id, a.output as output, sg.id as sequencing_group_id
+SELECT peid.external_id as participant_id, a.output as output, sg.id as sequencing_group_id
 FROM analysis a
 INNER JOIN analysis_sequencing_group a_sg ON a_sg.analysis_id = a.id
 INNER JOIN sequencing_group sg ON a_sg.sequencing_group_id = sg.id
 INNER JOIN sample s ON sg.sample_id = s.id
-INNER JOIN participant p ON s.participant_id = p.id
+INNER JOIN participant_external_id peid ON s.participant_id = peid.participant_id
 WHERE
     {' AND '.join(filters)}
 ORDER BY a.timestamp_completed DESC;

deploy/python/version.txt

@@ -1 +1 @@
-7.1.0
+7.1.1

setup.py

@@ -19,7 +19,7 @@
 setup(
     name=PKG,
     # This tag is automatically updated by bump2version
-    version='7.1.0',
+    version='7.1.1',
     description='Python API for interacting with the Sample API system',
     long_description=readme,
     long_description_content_type='text/markdown',

test/test_analysis.py

@@ -3,6 +3,7 @@
 
 from db.python.layers.analysis import AnalysisLayer
 from db.python.layers.assay import AssayLayer
+from db.python.layers.participant import ParticipantLayer
 from db.python.layers.sample import SampleLayer
 from db.python.layers.sequencing_group import SequencingGroupLayer
 from db.python.tables.analysis import AnalysisFilter
@@ -12,6 +13,7 @@
     PRIMARY_EXTERNAL_ORG,
     AnalysisInternal,
     AssayUpsertInternal,
+    ParticipantUpsertInternal,
     SampleUpsertInternal,
     SequencingGroupUpsertInternal,
 )
@@ -20,6 +22,8 @@
 class TestAnalysis(DbIsolatedTest):
     """Test sample class"""
 
+    # pylint: disable=too-many-instance-attributes
+
     @run_as_sync
     async def setUp(self) -> None:
         # don't need to await because it's tagged @run_as_sync
@@ -28,6 +32,7 @@ async def setUp(self) -> None:
         self.sgl = SequencingGroupLayer(self.connection)
         self.asl = AssayLayer(self.connection)
         self.al = AnalysisLayer(self.connection)
+        self.pl = ParticipantLayer(self.connection)
 
         sample = await self.sl.upsert_sample(
             SampleUpsertInternal(
@@ -167,3 +172,22 @@ async def test_get_analysis(self):
         ]
 
         self.assertEqual(analyses, expected)
+
+    @run_as_sync
+    async def test_get_sample_cram_path_map_for_seqr(self):
+        """
+        Exercise get_sample_cram_path_map_for_seqr()
+        """
+
+        part = await self.pl.upsert_participants(
+            [
+                ParticipantUpsertInternal(
+                    external_ids={PRIMARY_EXTERNAL_ORG: 'PEXT1'},
+                    meta={},
+                    samples=[SampleUpsertInternal(id=self.sample_id)],
+                ),
+            ],
+        )
+
+        id_map = await self.al.get_sample_cram_path_map_for_seqr(self.project_id, ['blood'], [part[0].id])
+        self.assertIsInstance(id_map, list)

web/package.json

@@ -1,6 +1,6 @@
 {
     "name": "metamist",
-    "version": "7.1.0",
+    "version": "7.1.1",
     "private": true,
     "dependencies": {
         "@apollo/client": "^3.7.3",