format

Author: bpblanken

v03_pipeline/lib/tasks/update_lookup_table.py

@@ -1,14 +1,15 @@
 import hail as hl
 import luigi
+import luigi.util
 
 from v03_pipeline.lib.misc.callsets import callset_project_pairs
 from v03_pipeline.lib.misc.lookup import (
     compute_callset_lookup_ht,
     join_lookup_hts,
     remove_family_guids,
 )
-from v03_pipeline.lib.model import SampleType
 from v03_pipeline.lib.model.constants import PROJECTS_EXCLUDED_FROM_LOOKUP
+from v03_pipeline.lib.tasks.base.base_loading_run_params import BaseLoadingRunParams
 from v03_pipeline.lib.tasks.base.base_update_lookup_table import (
     BaseUpdateLookupTableTask,
 )
@@ -17,25 +18,11 @@
 )
 
 
+@luigi.util.inherits(BaseLoadingRunParams)
 class UpdateLookupTableTask(BaseUpdateLookupTableTask):
-    sample_type = luigi.EnumParameter(enum=SampleType)
-    callset_paths = luigi.ListParameter()
     project_guids = luigi.ListParameter()
     project_remap_paths = luigi.ListParameter()
     project_pedigree_paths = luigi.ListParameter()
-    imputed_sex_paths = luigi.ListParameter(default=None)
-    ignore_missing_samples_when_remapping = luigi.BoolParameter(
-        default=False,
-        parsing=luigi.BoolParameter.EXPLICIT_PARSING,
-    )
-    validate = luigi.BoolParameter(
-        default=True,
-        parsing=luigi.BoolParameter.EXPLICIT_PARSING,
-    )
-    force = luigi.BoolParameter(
-        default=False,
-        parsing=luigi.BoolParameter.EXPLICIT_PARSING,
-    )
 
     def complete(self) -> bool:
         return (
@@ -47,23 +34,11 @@ def complete(self) -> bool:
                         [
                             updates.contains(
                                 hl.Struct(
-                                    callset=callset_path,
+                                    callset=self.callset_path,
                                     project_guid=project_guid,
                                 ),
                             )
-                            for (
-                                callset_path,
-                                project_guid,
-                                _,
-                                _,
-                                _,
-                            ) in callset_project_pairs(
-                                self.callset_paths,
-                                self.project_guids,
-                                self.project_remap_paths,
-                                self.project_pedigree_paths,
-                                self.imputed_sex_paths,
-                            )
+                            for project_guid in self.project_guids
                         ],
                     ),
                     hl.read_table(self.output().path).updates,
@@ -73,31 +48,21 @@ def complete(self) -> bool:
 
     def requires(self) -> list[luigi.Task]:
         return [
-            WriteRemappedAndSubsettedCallsetTask(
-                self.reference_genome,
-                self.dataset_type,
-                self.sample_type,
-                callset_path,
-                project_guid,
-                project_remap_path,
-                project_pedigree_path,
-                imputed_sex_path,
-                self.ignore_missing_samples_when_remapping,
-                self.validate,
-                False,
+            self.clone(
+                WriteRemappedAndSubsettedCallsetTask,
+                project_guid=project_guid,
+                project_remap_path=project_remap_path,
+                project_pedigree_path=project_pedigree_path,
             )
             for (
-                callset_path,
                 project_guid,
                 project_remap_path,
                 project_pedigree_path,
-                imputed_sex_path,
-            ) in callset_project_pairs(
-                self.callset_paths,
+            ) in zip(
                 self.project_guids,
                 self.project_remap_paths,
                 self.project_pedigree_paths,
-                self.imputed_sex_paths,
+                strict=False,
             )
         ]
 

v03_pipeline/lib/tasks/update_lookup_table_test.py

@@ -19,7 +19,7 @@ def test_skip_update_lookup_table_task(self) -> None:
             reference_genome=ReferenceGenome.GRCh38,
             dataset_type=DatasetType.SNV_INDEL,
             sample_type=SampleType.WGS,
-            callset_paths=[TEST_VCF],
+            callset_path=TEST_VCF,
             project_guids=[
                 'R0555_seqr_demo',
             ],  # a project excluded from the lookup table
@@ -52,7 +52,7 @@ def test_update_lookup_table_task(self) -> None:
             reference_genome=ReferenceGenome.GRCh38,
             dataset_type=DatasetType.SNV_INDEL,
             sample_type=SampleType.WGS,
-            callset_paths=[TEST_VCF],
+            callset_path=TEST_VCF,
             project_guids=['R0113_test_project'],
             project_remap_paths=[TEST_REMAP],
             project_pedigree_paths=[TEST_PEDIGREE_3],

v03_pipeline/lib/tasks/update_project_table.py

@@ -8,39 +8,19 @@
     join_family_entries_hts,
     remove_family_guids,
 )
-from v03_pipeline.lib.model import SampleType
+from v03_pipeline.lib.tasks.base.base_loading_run_params import BaseLoadingRunParams
 from v03_pipeline.lib.tasks.base.base_update_project_table import (
     BaseUpdateProjectTableTask,
 )
 from v03_pipeline.lib.tasks.write_remapped_and_subsetted_callset import (
     WriteRemappedAndSubsettedCallsetTask,
 )
-from v03_pipeline.lib.tasks.base.base_loading_run_params import BaseLoadingRunParams
 
 
 @luigi.util.inherits(BaseLoadingRunParams)
 class UpdateProjectTableTask(BaseUpdateProjectTableTask):
-    sample_type = luigi.EnumParameter(enum=SampleType)
-    callset_path = luigi.Parameter()
     project_remap_path = luigi.Parameter()
     project_pedigree_path = luigi.Parameter()
-    imputed_sex_path = luigi.Parameter(default=None)
-    ignore_missing_samples_when_remapping = luigi.BoolParameter(
-        default=False,
-        parsing=luigi.BoolParameter.EXPLICIT_PARSING,
-    )
-    validate = luigi.BoolParameter(
-        default=True,
-        parsing=luigi.BoolParameter.EXPLICIT_PARSING,
-    )
-    force = luigi.BoolParameter(
-        default=False,
-        parsing=luigi.BoolParameter.EXPLICIT_PARSING,
-    )
-    is_new_gcnv_joint_call = luigi.BoolParameter(
-        default=False,
-        description='Is this a fully joint-called callset.',
-    )
 
     def complete(self) -> bool:
         return (
@@ -54,19 +34,7 @@ def complete(self) -> bool:
         )
 
     def requires(self) -> luigi.Task:
-        return WriteRemappedAndSubsettedCallsetTask(
-            self.reference_genome,
-            self.dataset_type,
-            self.sample_type,
-            self.callset_path,
-            self.project_guid,
-            self.project_remap_path,
-            self.project_pedigree_path,
-            self.imputed_sex_path,
-            self.ignore_missing_samples_when_remapping,
-            self.validate,
-            False,
-        )
+        return self.clone(WriteRemappedAndSubsettedCallsetTask, force=False)
 
     def update_table(self, ht: hl.Table) -> hl.Table:
         callset_mt = hl.read_matrix_table(self.input().path)