bugfix unsorted pedigree (#1029)

Author: bpblanken

v03_pipeline/lib/misc/pedigree.py

@@ -80,6 +80,9 @@ class Family:
     def __hash__(self):
         return hash(self.family_guid)
 
+    def __eq__(self, other):
+        return self.family_guid == other.family_guid
+
     @staticmethod
     def parse_direct_lineage(rows: list[hl.Struct]) -> dict[str, Sample]:
         samples = {}
@@ -162,7 +165,7 @@ def parse_pedigree_ht_to_families(
 ) -> set[Family]:
     families = set()
     for family_guid, rows in itertools.groupby(
-        pedigree_ht.collect(),
+        sorted(pedigree_ht.collect(), key=lambda x: x.family_guid),
         lambda x: x.family_guid,
     ):
         families.add(Family.parse(family_guid, list(rows)))

v03_pipeline/lib/misc/pedigree_test.py

@@ -8,6 +8,7 @@
 
 TEST_PEDIGREE_1 = 'v03_pipeline/var/test/pedigrees/test_pedigree_1.tsv'
 TEST_PEDIGREE_2 = 'v03_pipeline/var/test/pedigrees/test_pedigree_2.tsv'
+TEST_PEDIGREE_9 = 'v03_pipeline/var/test/pedigrees/test_pedigree_9.tsv'
 
 
 class PedigreesTest(unittest.TestCase):
@@ -490,3 +491,20 @@ def test_subsetted_pedigree_with_removed_parent(self) -> None:
             family.samples['BBL_BC1-000345_01_D1'].mother,
             'BBL_BC1-000345_03_D1',
         )
+
+    def test_pedigree_ungrouped_families(self) -> None:
+        pedigree_ht = import_pedigree(TEST_PEDIGREE_9)
+        parsed_pedigree = parse_pedigree_ht_to_families(pedigree_ht)
+        self.assertEqual(len(parsed_pedigree), 2)
+        family = next(
+            iter(
+                [
+                    family
+                    for family in parsed_pedigree
+                    if family.family_guid == 'family_2_1'
+                ],
+            ),
+        )
+        self.assertTrue(
+            family.samples.keys() == {'RGP_164_1', 'RGP_164_2', 'RGP_164_4'},
+        )

v03_pipeline/var/test/pedigrees/test_pedigree_9.tsv

@@ -0,0 +1,5 @@
+Project_GUID	Family_GUID	Family_ID	Individual_ID	Paternal_ID	Maternal_ID	Sex
+R0115_test_project2	family_2_1	family_2	RGP_164_1			F
+R0115_test_project2	family_2_1	family_2	RGP_164_2			M
+R0115_test_project2	family_2_3	family_3	RGP_165_1			F
+R0115_test_project2	family_2_1	family_2	RGP_164_4	RGP_164_1	RGP_164_2	F