pr cleanup

Author: bpblanken

v03_pipeline/lib/misc/family_entries.py

@@ -93,7 +93,7 @@ def deglobalize_ids(ht: hl.Table) -> hl.Table:
     return ht.drop('family_guids', 'family_samples')
 
 
-def remove_new_callset_family_guids(
+def remove_family_guids(
     ht: hl.Table,
     family_guids: hl.SetExpression,
 ) -> hl.Table:

v03_pipeline/lib/misc/family_entries_test.py

@@ -7,7 +7,7 @@
     deglobalize_ids,
     globalize_ids,
     join_family_entries_hts,
-    remove_new_callset_family_guids,
+    remove_family_guids,
 )
 from v03_pipeline.lib.model import DatasetType
 
@@ -202,7 +202,7 @@ def test_globalize_and_deglobalize(self) -> None:
             ],
         )
 
-    def test_remove_new_callset_family_guids(self) -> None:
+    def test_remove_family_guids(self) -> None:
         family_entries_ht = hl.Table.parallelize(
             [
                 {
@@ -244,7 +244,7 @@ def test_remove_new_callset_family_guids(self) -> None:
                 },
             ),
         )
-        family_entries_ht = remove_new_callset_family_guids(
+        family_entries_ht = remove_family_guids(
             family_entries_ht,
             hl.set(['012']),
         )
@@ -329,7 +329,7 @@ def test_remove_new_callset_family_guids_all_families(self) -> None:
                 },
             ),
         )
-        ht = remove_new_callset_family_guids(family_entries_ht, hl.set(['012', '123']))
+        ht = remove_family_guids(family_entries_ht, hl.set(['012', '123']))
         self.assertCountEqual(
             ht.globals.collect(),
             [hl.Struct(family_guids=[], family_samples={})],

v03_pipeline/lib/misc/lookup.py

@@ -59,7 +59,7 @@ def globalize_ids(ht: hl.Table, project_guid: str) -> hl.Table:
     )
 
 
-def remove_new_callset_family_guids(
+def remove_family_guids(
     ht: hl.Table,
     project_guid: str,
     family_guids: hl.SetExpression,
@@ -89,10 +89,10 @@ def remove_new_callset_family_guids(
     )
     return ht.annotate_globals(
         project_families=hl.dict(
-            ht.project_families.items().map(
-                lambda item: (
+            hl.enumerate(ht.project_families.items()).starmap(
+                lambda i, item: (
                     hl.if_else(
-                        item[0] != project_guid,
+                        i != project_i,
                         item,
                         (
                             item[0],

v03_pipeline/lib/misc/lookup_test.py

@@ -5,7 +5,7 @@
 from v03_pipeline.lib.misc.lookup import (
     compute_callset_lookup_ht,
     join_lookup_hts,
-    remove_new_callset_family_guids,
+    remove_family_guids,
 )
 from v03_pipeline.lib.model import DatasetType
 
@@ -168,17 +168,17 @@ def test_remove_new_callset_family_guids(self) -> None:
                 project_families={'project_a': ['1', '2', '3'], 'project_b': ['4']},
             ),
         )
-        lookup_ht = remove_new_callset_family_guids(
+        lookup_ht = remove_family_guids(
             lookup_ht,
             'project_c',
             hl.set(['2']),
         )
-        lookup_ht = remove_new_callset_family_guids(
+        lookup_ht = remove_family_guids(
             lookup_ht,
             'project_a',
             hl.set(['3', '1']),
         )
-        lookup_ht = remove_new_callset_family_guids(
+        lookup_ht = remove_family_guids(
             lookup_ht,
             'project_b',
             hl.set(['4']),

v03_pipeline/lib/tasks/update_lookup_table.py

@@ -4,7 +4,7 @@
 from v03_pipeline.lib.misc.lookup import (
     compute_callset_lookup_ht,
     join_lookup_hts,
-    remove_new_callset_family_guids,
+    remove_family_guids,
 )
 from v03_pipeline.lib.model.constants import PROJECTS_EXCLUDED_FROM_LOOKUP
 from v03_pipeline.lib.paths import lookup_table_path
@@ -101,7 +101,7 @@ def update_table(self, ht: hl.Table) -> hl.Table:
                 ),
             )
         callset_mt = hl.read_matrix_table(self.input().path)
-        ht = remove_new_callset_family_guids(
+        ht = remove_family_guids(
             ht,
             self.project_guid,
             callset_mt.index_globals().family_samples.key_set(),

v03_pipeline/lib/tasks/update_project_table.py

@@ -5,7 +5,7 @@
 from v03_pipeline.lib.misc.family_entries import (
     compute_callset_family_entries_ht,
     join_family_entries_hts,
-    remove_new_callset_family_guids,
+    remove_family_guids,
 )
 from v03_pipeline.lib.paths import project_table_path
 from v03_pipeline.lib.tasks.base.base_update_task import BaseUpdateTask
@@ -104,7 +104,7 @@ def update_table(self, ht: hl.Table) -> hl.Table:
                     callset_ht.family_entries.dtype.element_type,
                 ),
             )
-        ht = remove_new_callset_family_guids(
+        ht = remove_family_guids(
             ht,
             callset_mt.index_globals().family_samples.key_set(),
         )