reformat filters annotation

Author: bpblanken

v03_pipeline/lib/misc/io.py

@@ -121,7 +121,6 @@ def import_callset(
     callset_path: str,
     reference_genome: ReferenceGenome,
     dataset_type: DatasetType,
-    filters_path: str | None = None,
 ) -> hl.MatrixTable:
     if dataset_type == DatasetType.GCNV:
         mt = import_gcnv_bed_file(callset_path)
@@ -131,9 +130,6 @@ def import_callset(
         mt = hl.read_matrix_table(callset_path)
     if dataset_type == DatasetType.SV:
         mt = mt.annotate_rows(variant_id=mt.rsid)
-    if filters_path:
-        filters_ht = import_vcf(filters_path, reference_genome).rows()
-        mt = mt.annotate_rows(filters=filters_ht[mt.row_key].filters)
     return mt.key_rows_by(*dataset_type.table_key_type(reference_genome).fields)
 
 

v03_pipeline/lib/tasks/write_imported_callset.py

@@ -4,6 +4,7 @@
 
 from v03_pipeline.lib.misc.io import (
     import_callset,
+    import_vcf,
     select_relevant_fields,
     split_multi_hts,
 )
@@ -117,17 +118,22 @@ def additional_row_fields(self, mt):
         }
 
     def create_table(self) -> hl.MatrixTable:
-        filters_path = valid_filters_path(
-            self.dataset_type,
-            self.sample_type,
-            self.callset_path,
-        )
         mt = import_callset(
             self.callset_path,
             self.reference_genome,
             self.dataset_type,
-            filters_path,
         )
+        filters_path = None
+        if not self.skip_expect_filters and self.dataset_type.expect_filters(
+            self.sample_type,
+        ):
+            filters_path = valid_filters_path(
+                self.dataset_type,
+                self.sample_type,
+                self.callset_path,
+            )
+            filters_ht = import_vcf(filters_path, self.reference_genome).rows()
+            mt = mt.annotate_rows(filters=filters_ht[mt.row_key].filters)
         mt = select_relevant_fields(
             mt,
             self.dataset_type,