Merge branch 'main' of github.com:broadinstitute/seqr-loading-pipelines

Author: bpblanken

v03_pipeline/api/app_test.py

@@ -89,7 +89,6 @@ async def test_loading_pipeline_enqueue(self):
                 'Successfully queued': {
                     'callset_path': 'v03_pipeline/var/test/callsets/1kg_30variants.vcf',
                     'dataset_type': 'SNV_INDEL',
-                    'ignore_missing_samples_when_remapping': False,
                     'projects_to_run': ['project_a'],
                     'reference_genome': 'GRCh38',
                     'sample_type': 'WGS',

v03_pipeline/api/model.py

@@ -12,7 +12,6 @@ class LoadingPipelineRequest(BaseModel):
     sample_type: SampleType
     reference_genome: ReferenceGenome
     dataset_type: DatasetType
-    ignore_missing_samples_when_remapping: bool = False
     skip_validation: bool = False
     skip_check_sex_and_relatedness: bool = False
 

v03_pipeline/lib/misc/sample_ids.py

@@ -11,7 +11,6 @@
 def remap_sample_ids(
     mt: hl.MatrixTable,
     project_remap_ht: hl.Table,
-    ignore_missing_samples_when_remapping: bool,
 ) -> hl.MatrixTable:
     collected_remap = project_remap_ht.collect()
     s_dups = [k for k, v in Counter([r.s for r in collected_remap]).items() if v > 1]
@@ -31,12 +30,8 @@ def remap_sample_ids(
             f'Only {project_remap_ht.semi_join(mt.cols()).count()} out of {remap_count} '
             'remap IDs matched IDs in the variant callset.\n'
             f"IDs that aren't in the callset: {missing_samples}\n"
-            f'All callset sample IDs:{mt.s.collect()}'
         )
-        if ignore_missing_samples_when_remapping:
-            logger.info(message)
-        else:
-            raise SeqrValidationError(message)
+        raise SeqrValidationError(message)
 
     mt = mt.annotate_cols(**project_remap_ht[mt.s])
     remap_expr = hl.if_else(hl.is_missing(mt.seqr_id), mt.s, mt.seqr_id)

v03_pipeline/lib/misc/sample_ids_test.py

@@ -38,7 +38,6 @@ def test_remap_2_sample_ids(self) -> None:
         remapped_mt = remap_sample_ids(
             CALLSET_MT,
             project_remap_ht,
-            ignore_missing_samples_when_remapping=True,
         )
 
         self.assertEqual(remapped_mt.cols().count(), 3)
@@ -80,7 +79,6 @@ def test_remap_sample_ids_remap_has_duplicate(self) -> None:
             remap_sample_ids(
                 CALLSET_MT,
                 project_remap_ht,
-                ignore_missing_samples_when_remapping=True,
             )
 
     def test_remap_sample_ids_remap_has_missing_samples(self) -> None:
@@ -103,7 +101,6 @@ def test_remap_sample_ids_remap_has_missing_samples(self) -> None:
             remap_sample_ids(
                 CALLSET_MT,
                 project_remap_ht,
-                ignore_missing_samples_when_remapping=False,
             )
 
     def test_subset_samples_zero_samples(self):

v03_pipeline/lib/tasks/base/base_loading_run_params.py

@@ -22,10 +22,6 @@ class BaseLoadingRunParams(luigi.Task):
     project_guids = luigi.ListParameter(default=[])
     project_remap_paths = luigi.ListParameter(default=[])
     project_pedigree_paths = luigi.ListParameter(default=[])
-    ignore_missing_samples_when_remapping = luigi.BoolParameter(
-        default=False,
-        parsing=luigi.BoolParameter.EXPLICIT_PARSING,
-    )
     skip_check_sex_and_relatedness = luigi.BoolParameter(
         default=False,
         parsing=luigi.BoolParameter.EXPLICIT_PARSING,

v03_pipeline/lib/tasks/dataproc/misc_test.py

@@ -42,8 +42,6 @@ def test_to_kebab_str_args(self, _: Mock):
                 '["test_remap"]',
                 '--project-pedigree-paths',
                 '["test_pedigree"]',
-                '--ignore-missing-samples-when-remapping',
-                'False',
                 '--skip-check-sex-and-relatedness',
                 'False',
                 '--skip-expect-filters',

v03_pipeline/lib/tasks/write_remapped_and_subsetted_callset.py

@@ -96,7 +96,6 @@ def create_table(self) -> hl.MatrixTable:
             callset_mt = remap_sample_ids(
                 callset_mt,
                 project_remap_ht,
-                self.ignore_missing_samples_when_remapping,
             )
             remap_lookup = hl.dict(
                 {r.s: r.seqr_id for r in project_remap_ht.collect()},