Change local alleles default to false

Author: Will-Tyler

bio2zarr/cli.py

@@ -152,7 +152,7 @@ def list_commands(self, ctx):
 local_alleles = click.option(
     "--local-alleles/--no-local-alleles",
     show_default=True,
-    default=True,
+    default=False,
     help="Use local allele fields to reduce the storage requirements of the output.",
 )
 

bio2zarr/vcf2zarr/icf.py

@@ -1155,7 +1155,7 @@ def init(
         if compressor is None:
             compressor = ICF_DEFAULT_COMPRESSOR
         if local_alleles is None:
-            local_alleles = True
+            local_alleles = False
         vcfs = [pathlib.Path(vcf) for vcf in vcfs]
         target_num_partitions = max(target_num_partitions, len(vcfs))
 

bio2zarr/vcf2zarr/verification.py

@@ -170,10 +170,6 @@ def verify(vcf_path, zarr_path, show_progress=False):
     for colname in root.keys():
         if colname.startswith("call") and not colname.startswith("call_genotype"):
             vcf_name = colname.split("_", 1)[1]
-            if vcf_name == "LAA" and vcf_name not in format_headers:
-                continue  # LAA could have been computed during the explode step.
-            if vcf_name == "LPL" and vcf_name not in format_headers:
-                continue  # LPL could have been computed during the explode step.
             vcf_type = format_headers[vcf_name]["Type"]
             vcf_number = format_headers[vcf_name]["Number"]
             format_fields[vcf_name] = vcf_type, vcf_number, iter(root[colname])

tests/test_cli.py

@@ -14,7 +14,7 @@
     compressor=None,
     worker_processes=1,
     show_progress=True,
-    local_alleles=True,
+    local_alleles=False,
 )
 
 DEFAULT_DEXPLODE_PARTITION_ARGS = dict()
@@ -24,7 +24,7 @@
     column_chunk_size=64,
     compressor=None,
     show_progress=True,
-    local_alleles=True,
+    local_alleles=False,
 )
 
 DEFAULT_ENCODE_ARGS = dict(
@@ -54,7 +54,7 @@
     samples_chunk_size=None,
     show_progress=True,
     worker_processes=1,
-    local_alleles=True,
+    local_alleles=False,
 )
 
 DEFAULT_PLINK_CONVERT_ARGS = dict(

tests/test_vcf_examples.py

@@ -537,7 +537,7 @@ class Test1000G2020Example:
     @pytest.fixture(scope="class")
     def ds(self, tmp_path_factory):
         out = tmp_path_factory.mktemp("data") / "example.vcf.zarr"
-        vcf2zarr.convert([self.data_path], out, worker_processes=0)
+        vcf2zarr.convert([self.data_path], out, worker_processes=0, local_alleles=True)
         return sg.load_dataset(out)
 
     def test_position(self, ds):