Merge pull request #5048 from broadinstitute/airtable-remap-issue

use correct field for querying sample remap

Author: hanars

seqr/views/apis/data_manager_api.py

@@ -389,8 +389,8 @@ def _get_valid_search_individuals(project, airtable_samples, vcf_samples, datase
         }
         def fetch_missing_vcf_samples(missing_vcf_samples):
             samples = _get_dataset_type_samples_for_matched_pdos(
-                LOADABLE_PDO_STATUSES + AVAILABLE_PDO_STATUSES, **get_sample_kwargs,
-                additional_sample_filters={'VCFIDWithMismatch': sorted(missing_vcf_samples)},
+                LOADABLE_PDO_STATUSES + AVAILABLE_PDO_STATUSES, **get_sample_kwargs, sample_fields=['VCFIDWithMismatch'],
+                additional_sample_filters={'SeqrIDWithMismatch': sorted(missing_vcf_samples)},
             )
             vcf_sample_id_map.update({
                 s['sample_id']: s['VCFIDWithMismatch'] for s in samples

seqr/views/apis/data_manager_api_tests.py

@@ -1889,8 +1889,8 @@ def _assert_expected_airtable_call(self, required_sample_field, project_guid, ca
 
     def _assert_expected_airtable_vcf_id_call(self, required_sample_field=None, additional_vcf_ids='', **kwargs):
         self._assert_expected_airtable_call(
-            required_sample_field, 'R0004_non_analyst_project', **kwargs, additional_fields=['VCFIDWithMismatch'],
-            additional_filter=f"OR(VCFIDWithMismatch='NA21234'{additional_vcf_ids})",
+            required_sample_field, 'R0004_non_analyst_project', **kwargs, additional_fields=['VCFIDWithMismatch', 'SeqrIDWithMismatch'],
+            additional_filter=f"OR(SeqrIDWithMismatch='NA21234'{additional_vcf_ids})",
             additional_pdo_statuses=",SEARCH('Methods (Loading)',ARRAYJOIN(PDOStatus,';')),SEARCH('On hold for phenotips, but ready to load',ARRAYJOIN(PDOStatus,';'))",
         )
 
@@ -1944,7 +1944,7 @@ def _trigger_error(self, url, body, variables, mock_open, mock_gzip_open, mock_m
         body['sampleType'] = 'WGS'
         self.assertEqual(len(responses.calls), 1)
         self._assert_expected_airtable_vcf_id_call(
-            required_sample_field='gCNV_CallsetPath', additional_vcf_ids=",VCFIDWithMismatch='NA21987'",
+            required_sample_field='gCNV_CallsetPath', additional_vcf_ids=",SeqrIDWithMismatch='NA21987'",
         )
 
         responses.calls.reset()

seqr/views/utils/airtable_utils.py

@@ -171,7 +171,7 @@ def get_samples_for_sample_ids(self, sample_ids, fields):
             records_by_id.update(self._get_samples_for_id_field(missing, 'SeqrCollaboratorSampleID', fields))
         return records_by_id
 
-    def get_samples_for_matched_pdos(self, pdo_statuses, pdo_fields=None, additional_sample_filters=None, project_guid=None, required_sample_fields=None):
+    def get_samples_for_matched_pdos(self, pdo_statuses, pdo_fields=None, sample_fields=None, additional_sample_filters=None, project_guid=None, required_sample_fields=None):
         pdo_fields = pdo_fields or []
         additional_and_filters = [
             f'LEN({{{required_sample_field}}})>0' for required_sample_field in required_sample_fields or []
@@ -182,7 +182,7 @@ def get_samples_for_matched_pdos(self, pdo_statuses, pdo_fields=None, additional
         sample_records = self.fetch_records(
             'Samples', fields=[
                 'CollaboratorSampleID', 'SeqrCollaboratorSampleID', 'PDOStatus', 'SeqrProject', *pdo_fields,
-                *(additional_sample_filters or {}).keys(),
+                *(sample_fields or []), *(additional_sample_filters or {}).keys(),
             ],
             or_filters={'PDOStatus': pdo_statuses},
             and_filters={'SeqrProject': f'{BASE_URL}project/{project_guid}/project_page'} if project_guid else {},