fix outlier manage tests

seqr/management/commands/load_rna_seq.py

@@ -38,4 +38,4 @@ def _save_sample_data(self, sample_guid, data_by_gene):
         sample = Sample.objects.get(guid=sample_guid)
         models = self.model_cls.objects.bulk_create(
             [self.model_cls(sample=sample, **data) for data in data_by_gene.values()], batch_size=1000)
-        logger.info(f'create {len(models)} {self.model_cls} for {sample.sample_id}')
+        logger.info(f'create {len(models)} {self.model_cls.__name__} for {sample.sample_id}')

seqr/management/tests/load_rna_seq_tests.py

@@ -96,4 +96,58 @@ def test_tpm(self, mock_open, mock_logger, mock_utils_logger, mock_gzip_open):
         self.assertEqual(models.values('sample').distinct().count(), 2)
         mock_logger.info.assert_has_calls([
             mock.call('create 1 RnaSeqTpm for NA19678_D1'),
+            mock.call('DONE'),
+        ])
+
+    @mock.patch('seqr.management.commands.load_rna_seq.logger.info')
+    @mock.patch('seqr.management.commands.load_rna_seq.open')
+    @mock.patch('seqr.utils.file_utils.gzip.open')
+    def test_outlier(self, mock_gzip_open, mock_open, mock_logger):
+        mock_gzip_file = mock_gzip_open.return_value.__enter__.return_value
+        mock_gzip_file.__iter__.return_value = ['invalid\theader']
+
+        with self.assertRaises(ValueError) as e:
+            call_command('load_rna_seq', 'outlier', RNA_FILE_ID)
+        self.assertEqual(str(e.exception),
+                         'Invalid file: missing column(s): geneID, pValue, padjust, project, sampleID, tissue, zScore')
+
+        mock_gzip_file.__iter__.return_value = [
+            'sampleID\tproject\tgeneID\tdetail\tpValue\tpadjust\tzScore\ttissue\n',
+            'NA19675_D2\t1kg project nåme with uniçøde\tENSG00000240361\tdetail1\t0.01\t0.13\t-3.1\tmuscle\n',
+            'NA19675_D2\t1kg project nåme with uniçøde\tENSG00000240361\tdetail2\t0.01\t0.13\t-3.1\tmuscle\n',
+            'NA19675_D2\t1kg project nåme with uniçøde\tENSG00000233750\tdetail1\t0.064\t0.0000057\t7.8\tmuscle\n',
+            'NA19675_D3\t1kg project nåme with uniçøde\tENSG00000233750\tdetail1\t0.064\t0.0000057\t7.8\tmuscle\n',
+            'NA19675_D4\t1kg project nåme with uniçøde\tENSG00000233750\tdetail1\t0.064\t0.0000057\t7.8\tmuscle\n',
+        ]
+        mock_open.return_value.__enter__.return_value.__iter__.return_value = ['NA19675_D4\tNA19678']
+
+        with self.assertRaises(ErrorsWarningsException) as e:
+            call_command('load_rna_seq', 'outlier', RNA_FILE_ID)
+        self.assertEqual(e.exception.errors,
+                         ['Unable to find matches for the following samples: NA19675_D3, NA19675_D4'])
+
+        with self.assertRaises(ErrorsWarningsException) as e:
+            call_command('load_rna_seq', 'outlier', RNA_FILE_ID, '--mapping-file', 'map.tsv')
+        self.assertEqual(e.exception.errors, ['Unable to find matches for the following samples: NA19675_D3'])
+
+        call_command('load_rna_seq', 'outlier', RNA_FILE_ID, '--ignore-extra-samples')
+
+        rna_samples = Sample.objects.filter(individual_id=1, sample_id='NA19675_D2', sample_type='RNA')
+        self.assertEqual(len(rna_samples), 1)
+        sample = rna_samples.first()
+        self.assertEqual(sample.guid, EXISTING_SAMPLE_GUID)
+        self.assertTrue(sample.is_active)
+        self.assertIsNone(sample.elasticsearch_index)
+        #self.assertEqual(sample.data_source, 'new_muscle_samples.tsv.gz') TODO?
+        self.assertEqual(sample.tissue_type, 'M')
+
+        models = RnaSeqOutlier.objects.all()
+        self.assertEqual(models.count(), 2)
+        self.assertSetEqual({model.sample for model in models}, {sample})
+        self.assertListEqual(list(models.values_list('gene_id', 'p_adjust', 'p_value', 'z_score')), [
+            ('ENSG00000240361', 0.13, 0.01, -3.1), ('ENSG00000233750', 0.0000057, 0.064, 7.8),
+        ])
+        mock_logger.assert_has_calls([
+            mock.call('create 2 RnaSeqOutlier for NA19675_D2'),
+            mock.call('DONE'),
         ])