cleanup

Author: bpblanken

v03_pipeline/lib/tasks/base/base_task.py renamed to v03_pipeline/lib/tasks/base/base_hail_table_task.py

@@ -7,7 +7,7 @@
 
 logger = get_logger(__name__)
 
-class BaseTask(luigi.Task):
+class BaseHailTableTask(luigi.Task):
     reference_genome = luigi.EnumParameter(enum=ReferenceGenome)
     dataset_type = luigi.EnumParameter(enum=DatasetType)
     sample_type = luigi.EnumParameter(enum=SampleType)
@@ -25,6 +25,8 @@ def init_hail(self):
         # Interval ref data join causes shuffle death, this prevents it
         hl._set_flags(use_new_shuffle='1', no_whole_stage_codegen='1')  # noqa: SLF001
 
+# NB: these are defined over luigi.Task instead of the BaseHailTableTask so that
+# they work on file dependencies.
 @luigi.Task.event_handler(luigi.Event.DEPENDENCY_MISSING)
 def dependency_missing(task):
     logger.info(f'{task} dependency_missing')

v03_pipeline/lib/tasks/base/base_update_task.py

@@ -1,10 +1,10 @@
 import hail as hl
 
 from v03_pipeline.lib.misc.io import write
-from v03_pipeline.lib.tasks.base.base_task import BaseTask
+from v03_pipeline.lib.tasks.base.base_hail_table_task import BaseHailTableTask
 
 
-class BaseUpdateTask(BaseTask):
+class BaseUpdateTask(BaseHailTableTask):
     def run(self) -> None:
         self.init_hail()
         if not self.output().exists():

v03_pipeline/lib/tasks/base/base_write_task.py

@@ -1,10 +1,10 @@
 import hail as hl
 
 from v03_pipeline.lib.misc.io import write
-from v03_pipeline.lib.tasks.base.base_task import BaseTask
+from v03_pipeline.lib.tasks.base.base_hail_table_task import BaseHailTableTask
 
 
-class BaseWriteTask(BaseTask):
+class BaseWriteTask(BaseHailTableTask):
     def run(self) -> None:
         self.init_hail()
         ht = self.create_table()