def test_header_header():
lines1 = [
header.HeaderLine("foo", "bar"),
header.HeaderLine("foo2", "bar2")
]
samples1 = header.SamplesInfos(["one", "two", "three"])
hdr1 = header.Header(lines1, samples1)
lines2 = [
header.HeaderLine("foo", "bar"),
header.HeaderLine("foo2", "bar2")
]
samples2 = header.SamplesInfos(["one", "two", "three"])
hdr2 = header.Header(lines2, samples2)
lines3 = [
header.HeaderLine("foo3", "bar"),
header.HeaderLine("foo2", "bar2")
]
samples3 = header.SamplesInfos(["one", "two", "three"])
hdr3 = header.Header(lines3, samples3)
assert hdr1 == hdr2
assert hdr1 != hdr3
EXPECTED = (
"Header(lines=[HeaderLine('foo', 'bar'), HeaderLine('foo2', 'bar2')], "
"samples=SamplesInfos(names=['one', 'two', 'three'], "
"name_to_idx={'one': 0, 'three': 2, 'two': 1}))")
assert str(hdr1) == EXPECTED
with pytest.raises(TypeError):
hash(hdr1)
def test_sample_infos():
info1 = header.SamplesInfos(["one", "two", "three"])
info2 = header.SamplesInfos(["one", "two", "three"])
info3 = header.SamplesInfos(["one", "two", "four"])
assert info1 == info2
assert info1 != info3
with pytest.raises(TypeError):
assert hash(info1)
assert (
str(info1) ==
"SamplesInfos(names=['one', 'two', 'three'], name_to_idx={'one': 0, 'three': 2, 'two': 1})"
)
assert (
repr(info1) ==
"SamplesInfos(names=['one', 'two', 'three'], name_to_idx={'one': 0, 'three': 2, 'two': 1})"
)
def test_header_has_header_line_positive_no_samples():
lines = []
samples = header.SamplesInfos(["one", "two", "three"])
hdr = header.Header(lines, samples)
assert not hdr.has_header_line("FORMAT", "DP")
assert not hdr.has_header_line("INFO", "AD")
assert not hdr.has_header_line("FILTER", "PASS")
assert not hdr.has_header_line("contig", "1")
def test_header_checker_missing_fileformat():
# construct Header
lines = [header.HeaderLine("key", "value")]
hdr = header.Header(lines=lines, samples=header.SamplesInfos(["NA001"]))
# setup HeaderChecker
stream = io.StringIO()
checker = parser.HeaderChecker()
# execute
with pytest.raises(exceptions.InvalidHeaderException):
checker.run(hdr)
def test_header_checker_known_vcf_version():
# construct Header
lines = [header.HeaderLine("fileformat", "VCFv4.3")]
hdr = header.Header(lines=lines, samples=header.SamplesInfos(["NA001"]))
# setup HeaderChecker
stream = io.StringIO()
checker = parser.HeaderChecker()
# execute
checker.run(hdr)
# check result
EXPECTED = ""
assert stream.getvalue() == EXPECTED
def test_header_without_lines():
lines = [
header.HeaderLine("foo", "bar"),
header.HeaderLine("foo2", "bar2")
]
samples = header.SamplesInfos(["one", "two", "three"])
hdr = header.Header(lines, samples)
hdr.add_filter_line(vcfpy.OrderedDict([("ID", "PASS")]))
hdr.add_filter_line(vcfpy.OrderedDict([("ID", "q30")]))
assert len(hdr.lines) == 4
hdr2 = header.header_without_lines(hdr, [("foo", "bar"),
("FILTER", "q30")])
assert len(hdr2.lines) == 2
assert hdr2.samples == hdr.samples
def test_write_header_no_samples(tmpdir_factory):
# Create header to write out from scratch
hdr = header.Header(
lines=[header.HeaderLine("fileformat", "VCFv4.0")], samples=header.SamplesInfos([])
)
# Write out header
path = tmpdir_factory.mktemp("write_header").join("out.vcf")
w = writer.Writer.from_path(path, hdr)
w.close()
# Compare result
RESULT = path.read()
EXPECTED = textwrap.dedent(
r"""
##fileformat=VCFv4.0
#CHROM POS ID REF ALT QUAL FILTER INFO
"""
).lstrip()
assert RESULT == EXPECTED
def test_write_record_no_samples(tmpdir_factory):
O = vcfpy.OrderedDict
# Create header without samples
hdr = header.Header(lines=[header.HeaderLine("fileformat", "VCFv4.0")],
samples=header.SamplesInfos([]))
# construct record to write out from scratch
r = record.Record("20", 100, [], "C",
[record.Substitution(record.SNV, "T")], None, [], O())
# Write out header and record
path = tmpdir_factory.mktemp("write_header").join("out.vcf")
w = writer.Writer.from_path(path, hdr)
w.write_record(r)
w.close()
# Compare result
RESULT = path.read()
EXPECTED = textwrap.dedent("""
##fileformat=VCFv4.0
#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO
20\t100\t.\tC\tT\t.\t.\t.
""").lstrip()
assert RESULT == EXPECTED
def test_header_has_header_line_positive():
lines = [
header.FormatHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "DP"), ("Number", "R"),
("Type", "Integer")])),
header.InfoHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "AD"), ("Number", "R"),
("Type", "Integer")])),
header.FilterHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "PASS"),
("Description", "All filters passed")])),
header.ContigHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "1"), ("length", 234)])),
]
samples = header.SamplesInfos(["one", "two", "three"])
hdr = header.Header(lines, samples)
assert hdr.has_header_line("FORMAT", "DP")
assert hdr.has_header_line("INFO", "AD")
assert hdr.has_header_line("FILTER", "PASS")
assert hdr.has_header_line("contig", "1")
