From a30a41a33425d0a9f8873e745b925b219441aaa7 Mon Sep 17 00:00:00 2001
From: Paul Pillot <paul.pillot@mcgill.ca>
Date: Tue, 19 Apr 2022 10:49:12 -0400
Subject: [PATCH] skip samples with no individual

---
 chord_metadata_service/chord/export_cbio.py   |  4 ++++
 .../chord/tests/test_export_cbio.py           | 23 +++++++++++++++----
 2 files changed, 23 insertions(+), 4 deletions(-)

diff --git a/chord_metadata_service/chord/export_cbio.py b/chord_metadata_service/chord/export_cbio.py
index 95920b055..20788298f 100644
--- a/chord_metadata_service/chord/export_cbio.py
+++ b/chord_metadata_service/chord/export_cbio.py
@@ -175,6 +175,10 @@ def sample_export(results, file_handle: TextIO):
 
     samples = []
     for sample in results:
+        # sample.inidividual can be null. Skip the sample in that case.
+        if sample.individual is None:
+            continue
+
         sample_obj = {
             'individual_id': sample.individual.id,
             'id': sample.id
diff --git a/chord_metadata_service/chord/tests/test_export_cbio.py b/chord_metadata_service/chord/tests/test_export_cbio.py
index b3c74ca6d..4e9fddf78 100644
--- a/chord_metadata_service/chord/tests/test_export_cbio.py
+++ b/chord_metadata_service/chord/tests/test_export_cbio.py
@@ -58,6 +58,11 @@ def setUp(self) -> None:
 
         self.p = WORKFLOW_INGEST_FUNCTION_MAP[WORKFLOW_PHENOPACKETS_JSON](EXAMPLE_INGEST_OUTPUTS, self.t.identifier)
 
+        # Update the last sample to remove reference to any individual.
+        PhModel.Biosample.objects.filter(
+                id=EXAMPLE_INGEST_PHENOPACKET["biosamples"][-1]["id"]
+            ).update(individual=None)
+
     def stream_to_dict(self, output: TextIO) -> Dict[str, str]:
         """
         Utility function. Parses cBioPortal meta data text files (lines of
@@ -158,6 +163,7 @@ def test_export_cbio_sample_data(self):
             output.seek(0)
             field_count = None
             field_names = []
+            sample_count = 0
             for i, line in enumerate(output):
                 # 4 first header lines begin with `#`
                 if i < 4:
@@ -178,10 +184,19 @@ def test_export_cbio_sample_data(self):
                     self.assertIn('SAMPLE_ID', pieces)
                     continue
 
-                # TSV body. Inspect first line and break
+                # TSV body.
                 self.assertEqual(field_count, len(pieces))
                 record = dict(zip(field_names, pieces))
 
-                self.assertEqual(record["PATIENT_ID"], EXAMPLE_INGEST_PHENOPACKET["subject"]["id"])
-                self.assertEqual(record["SAMPLE_ID"], EXAMPLE_INGEST_PHENOPACKET["biosamples"][0]["id"])
-                break
+                self.assertEqual(
+                    record["PATIENT_ID"],
+                    EXAMPLE_INGEST_PHENOPACKET["biosamples"][sample_count]["individual_id"]
+                )
+                self.assertEqual(
+                    record["SAMPLE_ID"],
+                    EXAMPLE_INGEST_PHENOPACKET["biosamples"][sample_count]["id"]
+                )
+                sample_count += 1
+
+            # samples not attached to an individual are not exported
+            self.assertEqual(sample_count, samples.filter(individual_id__isnull=False).count())