Loosen test for checking if VCF headers are the same

Author: tomwhite

tests/test_vcf_writer.py

@@ -6,7 +6,6 @@
 import numpy as np
 import pytest
 import zarr
-from bio2zarr import icf
 from cyvcf2 import VCF
 from numpy.testing import assert_array_equal
 
@@ -303,15 +302,9 @@ def test_write_vcf__header_flags(tmp_path):
     assert_vcfs_close(original, output)
 
 
-def test_write_vcf__generate_header(tmp_path):
+def test_write_vcf__generate_header():
     original = pathlib.Path("tests/data/vcf") / "sample.vcf.gz"
-    # don't use cache here since we mutate the vcz
-    vcz = tmp_path.joinpath("intermediate.vcz")
-    icf.convert([original], vcz, worker_processes=0, local_alleles=False)
-
-    # remove vcf_header
-    root = zarr.open(vcz, mode="r+")
-    del root.attrs["vcf_header"]
+    vcz = vcz_path_cache(original)
 
     output_header = StringIO()
     write_vcf(vcz, output_header, header_only=True, no_version=True)
@@ -326,9 +319,9 @@ def test_write_vcf__generate_header(tmp_path):
 ##INFO=<ID=DP,Number=1,Type=Integer,Description="Total Depth">
 ##INFO=<ID=H2,Number=0,Type=Flag,Description="HapMap2 membership">
 ##INFO=<ID=NS,Number=1,Type=Integer,Description="Number of Samples With Data">
-##FILTER=<ID=PASS,Description="">
-##FILTER=<ID=s50,Description="">
-##FILTER=<ID=q10,Description="">
+##FILTER=<ID=PASS,Description="All filters passed">
+##FILTER=<ID=s50,Description="Less than 50% of samples have data">
+##FILTER=<ID=q10,Description="Quality below 10">
 ##FORMAT=<ID=GT,Number=1,Type=String,Description="Genotype">
 ##FORMAT=<ID=DP,Number=1,Type=Integer,Description="Read Depth">
 ##FORMAT=<ID=GQ,Number=1,Type=Integer,Description="Genotype Quality">
@@ -340,6 +333,7 @@ def test_write_vcf__generate_header(tmp_path):
 """  # noqa: E501
 
     # substitute value of source
+    root = zarr.open(vcz, mode="r+")
     expected_vcf_header = expected_vcf_header.format(root.attrs["source"])
 
     assert output_header.getvalue() == expected_vcf_header

tests/utils.py

@@ -29,6 +29,45 @@ def normalise_info_missingness(info_dict, key):
     return value
 
 
+def _get_headers(vcf, header_type):
+    def to_dict(header_field):
+        d = header_field.info(extra=True)
+        del d[b"IDX"]  # remove IDX since we don't care about ordering
+
+        # cyvcf2 duplicates some keys as strings and bytes, so remove the bytes one
+        for k in list(d.keys()):
+            if isinstance(k, bytes) and k.decode("utf-8") in d:
+                del d[k]
+        return d
+
+    return {
+        field["ID"]: to_dict(field)
+        for field in vcf.header_iter()
+        if field["HeaderType"] == header_type
+    }
+
+
+def _assert_vcf_headers_equivalent(vcf1, vcf2):
+    # Only compare INFO, FORMAT, FILTER, CONTIG fields, ignoring order
+    # Other fields are ignored
+
+    info1 = _get_headers(vcf1, "INFO")
+    info2 = _get_headers(vcf2, "INFO")
+    assert info1 == info2
+
+    format1 = _get_headers(vcf1, "FORMAT")
+    format2 = _get_headers(vcf2, "FORMAT")
+    assert format1 == format2
+
+    filter1 = _get_headers(vcf1, "FILTER")
+    filter2 = _get_headers(vcf2, "FILTER")
+    assert filter1 == filter2
+
+    contig1 = _get_headers(vcf1, "CONTIG")
+    contig2 = _get_headers(vcf2, "CONTIG")
+    assert contig1 == contig2
+
+
 def assert_vcfs_close(f1, f2, *, rtol=1e-05, atol=1e-03):
     """Like :py:func:`numpy.testing.assert_allclose()`, but for VCF files.
 
@@ -48,7 +87,7 @@ def assert_vcfs_close(f1, f2, *, rtol=1e-05, atol=1e-03):
         Absolute tolerance.
     """
     with open_vcf(f1) as vcf1, open_vcf(f2) as vcf2:
-        assert vcf1.raw_header == vcf2.raw_header
+        _assert_vcf_headers_equivalent(vcf1, vcf2)
         assert vcf1.samples == vcf2.samples
 
         for v1, v2 in zip_longest(vcf1, vcf2):