def test_header_contig_header_line():
line1 = header.ContigHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "1"), ("length", 234)]))
line2 = header.ContigHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "1"), ("length", 234)]))
line3 = header.ContigHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "2"), ("length", 123)]))
assert line1 == line2
assert line1 != line3
if sys.version_info < (3, 6):
assert str(line1) == (
"ContigHeaderLine('contig', '<ID=1,length=234>', OrderedDict([('ID', '1'), ('length', 234)]))"
)
assert repr(line1) == (
"ContigHeaderLine('contig', '<ID=1,length=234>', OrderedDict([('ID', '1'), ('length', 234)]))"
)
else:
assert str(line1) == (
"ContigHeaderLine('contig', '<ID=1,length=234>', {'ID': '1', 'length': 234})"
)
assert repr(line1) == (
"ContigHeaderLine('contig', '<ID=1,length=234>', {'ID': '1', 'length': 234})"
)
assert line1.value == "<ID=1,length=234>"
assert line1.serialize() == "##contig=<ID=1,length=234>"
with pytest.raises(TypeError):
hash(line1)
def test_header_info_header_line():
line1 = header.InfoHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "SVTYPE"), ("Number", 1),
("Type", "String")]))
line2 = header.InfoHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "SVTYPE"), ("Number", 1),
("Type", "String")]))
line3 = header.InfoHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "END"), ("Number", 1), ("Type", "Integer")]))
assert line1 == line2
assert line1 != line3
if sys.version_info < (3, 6):
assert str(line1) == (
"InfoHeaderLine('INFO', '<ID=SVTYPE,Number=1,Type=String>', "
"OrderedDict([('ID', 'SVTYPE'), ('Number', 1), ('Type', 'String')]))"
)
assert repr(line1) == (
"InfoHeaderLine('INFO', '<ID=SVTYPE,Number=1,Type=String>', "
"OrderedDict([('ID', 'SVTYPE'), ('Number', 1), ('Type', 'String')]))"
)
else:
assert str(line1) == (
"InfoHeaderLine('INFO', '<ID=SVTYPE,Number=1,Type=String>', "
"{'ID': 'SVTYPE', 'Number': 1, 'Type': 'String'})")
assert repr(line1) == (
"InfoHeaderLine('INFO', '<ID=SVTYPE,Number=1,Type=String>', "
"{'ID': 'SVTYPE', 'Number': 1, 'Type': 'String'})")
assert line1.value == "<ID=SVTYPE,Number=1,Type=String>"
assert line1.serialize() == "##INFO=<ID=SVTYPE,Number=1,Type=String>"
with pytest.raises(TypeError):
hash(line1)
def test_header_format_header_line():
line1 = header.FormatHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "AD"), ("Number", "R"),
("Type", "Integer")]))
line2 = header.FormatHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "AD"), ("Number", "R"),
("Type", "Integer")]))
line3 = header.FormatHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "DP"), ("Number", 1), ("Type", "Integer")]))
assert line1 == line2
assert line1 != line3
if sys.version_info < (3, 6):
assert str(line1) == (
"FormatHeaderLine('FORMAT', '<ID=AD,Number=R,Type=Integer>', "
"OrderedDict([('ID', 'AD'), ('Number', 'R'), ('Type', 'Integer')]))"
)
assert repr(line1) == (
"FormatHeaderLine('FORMAT', '<ID=AD,Number=R,Type=Integer>', "
"OrderedDict([('ID', 'AD'), ('Number', 'R'), ('Type', 'Integer')]))"
)
else:
assert str(line1) == (
"FormatHeaderLine('FORMAT', '<ID=AD,Number=R,Type=Integer>', "
"{'ID': 'AD', 'Number': 'R', 'Type': 'Integer'})")
assert repr(line1) == (
"FormatHeaderLine('FORMAT', '<ID=AD,Number=R,Type=Integer>', "
"{'ID': 'AD', 'Number': 'R', 'Type': 'Integer'})")
assert line1.value == "<ID=AD,Number=R,Type=Integer>"
assert line1.serialize() == "##FORMAT=<ID=AD,Number=R,Type=Integer>"
with pytest.raises(TypeError):
hash(line1)
def test_header_sample_header_line():
line1 = header.SampleHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "sample1")]))
line2 = header.SampleHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "sample1")]))
line3 = header.SampleHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "sample2")]))
assert line1 == line2
assert line1 != line3
if sys.version_info < (3, 6):
assert str(line1) == (
"SampleHeaderLine('SAMPLE', '<ID=sample1>', OrderedDict([('ID', 'sample1')]))"
)
assert repr(line1) == (
"SampleHeaderLine('SAMPLE', '<ID=sample1>', OrderedDict([('ID', 'sample1')]))"
)
else:
assert str(line1) == (
"SampleHeaderLine('SAMPLE', '<ID=sample1>', {'ID': 'sample1'})")
assert repr(line1) == (
"SampleHeaderLine('SAMPLE', '<ID=sample1>', {'ID': 'sample1'})")
assert line1.value == "<ID=sample1>"
assert line1.serialize() == "##SAMPLE=<ID=sample1>"
with pytest.raises(TypeError):
hash(line1)
def test_header_pedigree_header_line():
line1 = header.PedigreeHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "child"), ("Father", "father")]))
line2 = header.PedigreeHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "child"), ("Father", "father")]))
line3 = header.PedigreeHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "father")]))
assert line1 == line2
assert line1 != line3
if sys.version_info < (3, 6):
assert str(line1) == (
"PedigreeHeaderLine('PEDIGREE', '<ID=child,Father=father>', "
"OrderedDict([('ID', 'child'), ('Father', 'father')]))")
assert repr(line1) == (
"PedigreeHeaderLine('PEDIGREE', '<ID=child,Father=father>', "
"OrderedDict([('ID', 'child'), ('Father', 'father')]))")
else:
assert str(line1) == (
"PedigreeHeaderLine('PEDIGREE', '<ID=child,Father=father>', {'ID': 'child', 'Father': 'father'})"
)
assert repr(line1) == (
"PedigreeHeaderLine('PEDIGREE', '<ID=child,Father=father>', {'ID': 'child', 'Father': 'father'})"
)
assert line1.value == "<ID=child,Father=father>"
assert line1.serialize() == "##PEDIGREE=<ID=child,Father=father>"
with pytest.raises(TypeError):
hash(line1)
def test_header_filter_header_line():
line1 = header.FilterHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "PASS"),
("Description", "All filters passed")]))
line2 = header.FilterHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "PASS"),
("Description", "All filters passed")]))
line3 = header.FilterHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "q30"), ("Description", "Phred score <30")]))
assert line1 == line2
assert line1 != line3
if sys.version_info < (3, 6):
assert str(line1) == (
"FilterHeaderLine('FILTER', '<ID=PASS,Description=\"All filters passed\">', "
"OrderedDict([('ID', 'PASS'), ('Description', 'All filters passed')]))"
)
assert repr(line1) == (
"FilterHeaderLine('FILTER', '<ID=PASS,Description=\"All filters passed\">', "
"OrderedDict([('ID', 'PASS'), ('Description', 'All filters passed')]))"
)
else:
assert str(line1) == (
"FilterHeaderLine('FILTER', '<ID=PASS,Description=\"All filters passed\">', "
"{'ID': 'PASS', 'Description': 'All filters passed'})")
assert repr(line1) == (
"FilterHeaderLine('FILTER', '<ID=PASS,Description=\"All filters passed\">', "
"{'ID': 'PASS', 'Description': 'All filters passed'})")
assert line1.value == '<ID=PASS,Description="All filters passed">'
assert line1.serialize(
) == '##FILTER=<ID=PASS,Description="All filters passed">'
with pytest.raises(TypeError):
hash(line1)
def test_header_alt_allele_header_line():
line1 = header.AltAlleleHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "DEL"), ("Description", "deletion")]))
line2 = header.AltAlleleHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "DEL"), ("Description", "deletion")]))
line3 = header.AltAlleleHeaderLine.from_mapping(
vcfpy.OrderedDict([("ID", "DUP"), ("Description", "duplication")]))
assert line1 == line2
assert line1 != line3
if sys.version_info < (3, 6):
assert str(line1) == (
"""AltAlleleHeaderLine('ALT', '<ID=DEL,Description="deletion">', """
"""OrderedDict([('ID', 'DEL'), ('Description', 'deletion')]))""")
assert repr(line1) == (
"""AltAlleleHeaderLine('ALT', '<ID=DEL,Description="deletion">', """
"""OrderedDict([('ID', 'DEL'), ('Description', 'deletion')]))""")
else:
assert str(line1) == (
"AltAlleleHeaderLine('ALT', '<ID=DEL,Description=\"deletion\">', "
"{'ID': 'DEL', 'Description': 'deletion'})")
assert repr(line1) == (
"AltAlleleHeaderLine('ALT', '<ID=DEL,Description=\"deletion\">', "
"{'ID': 'DEL', 'Description': 'deletion'})")
assert line1.value == '<ID=DEL,Description="deletion">'
assert line1.serialize() == '##ALT=<ID=DEL,Description="deletion">'
with pytest.raises(TypeError):
hash(line1)
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_add_contig_line_shortcut(vcf_header):
# check header before adding
assert len(vcf_header.lines) == 18
assert "20a" not in vcf_header._indices["contig"]
# add header line
mapping = vcfpy.OrderedDict([
("ID", "20a"),
("length", 62435964),
("assembly", "B36"),
("md5", "f126cdf8a6e0c7f379d618ff66beb2da"),
("species", "H**o sapiens"),
("taxonomy", "x"),
])
vcf_header.add_contig_line(mapping)
# check header after adding
assert len(vcf_header.lines) == 19
assert "20a" in vcf_header._indices["contig"]
assert vcf_header._indices["contig"]["20a"] is vcf_header.lines[-1]
# Check resulting added header line
assert vcf_header.lines[-1].key == "contig"
VALUE = ("<ID=20a,length=62435964,assembly=B36,"
"md5=f126cdf8a6e0c7f379d618ff66beb2da,"
'species="H**o sapiens",taxonomy=x>')
assert vcf_header.lines[-1].value == VALUE
assert len(vcf_header.lines[-1].mapping) == 6
assert vcf_header.lines[-1].mapping["ID"] == "20a"
assert vcf_header.lines[-1].mapping["length"] == 62435964
assert vcf_header.lines[-1].mapping["assembly"] == "B36"
assert vcf_header.lines[-1].mapping[
"md5"] == "f126cdf8a6e0c7f379d618ff66beb2da"
assert vcf_header.lines[-1].mapping["species"] == "H**o sapiens"
assert vcf_header.lines[-1].mapping["taxonomy"] == "x"
def test_add_info_line_shortcut(vcf_header):
# check header before adding
assert len(vcf_header.lines) == 18
# add header line
VALUE = '<ID=DPa,Number=1,Type=Integer,Description="Total Depth">'
mapping = vcfpy.OrderedDict([("ID", "DPa"), ("Number", 1),
("Type", "Integer"),
("Description", "Total Depth")])
vcf_header.add_info_line(mapping)
assert len(vcf_header.lines) == 19
# check header after adding
assert len(vcf_header.lines) == 19
assert "DPa" in vcf_header._indices["INFO"]
assert vcf_header._indices["INFO"]["DPa"] is vcf_header.lines[-1]
# Check resulting added header line
assert vcf_header.lines[-1].key == "INFO"
assert vcf_header.lines[-1].value == VALUE
assert len(vcf_header.lines[-1].mapping) == 4
assert vcf_header.lines[-1].mapping["ID"] == "DPa"
assert vcf_header.lines[-1].mapping["Number"] == 1
assert vcf_header.lines[-1].mapping["Type"] == "Integer"
assert vcf_header.lines[-1].mapping["Description"] == "Total Depth"
def test_add_format_line(vcf_header):
# check header before adding
assert len(vcf_header.lines) == 18
# add header line
VALUE = '<ID=GTa,Number=1,Type=String,Description="Genotype">'
line = header.FormatHeaderLine(
"FORMAT",
VALUE,
vcfpy.OrderedDict([("ID", "GTa"), ("Number", 1), ("Type", "String"),
("Description", "Genotype")]),
)
vcf_header.add_line(line)
# check header after adding
assert len(vcf_header.lines) == 19
assert "GTa" in vcf_header._indices["FORMAT"]
assert vcf_header._indices["FORMAT"]["GTa"] is vcf_header.lines[-1]
# Check resulting added header line
assert vcf_header.lines[-1].key == "FORMAT"
assert vcf_header.lines[-1].value == VALUE
assert len(vcf_header.lines[-1].mapping) == 4
assert vcf_header.lines[-1].mapping["ID"] == "GTa"
assert vcf_header.lines[-1].mapping["Number"] == 1
assert vcf_header.lines[-1].mapping["Type"] == "String"
assert vcf_header.lines[-1].mapping["Description"] == "Genotype"
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")
def build_rec(calls=None, format_extras=None):
calls = calls or []
format_extras = format_extras or []
alt1 = record.Substitution(vcfpy.SNV, "T")
alt2 = record.Substitution(vcfpy.SNV, "A")
return record.Record(
"2",
100,
[],
"C",
[alt1, alt2],
None,
[],
vcfpy.OrderedDict(),
["GT"] + format_extras,
calls,
)
def test_add_filter_line_shortcut(vcf_header):
# check header before adding
assert len(vcf_header.lines) == 18
# add header line
mapping = vcfpy.OrderedDict([("ID", "q10a"),
("Description", "Quality below 10")])
vcf_header.add_filter_line(mapping)
# check header after adding
assert len(vcf_header.lines) == 19
assert "q10a" in vcf_header._indices["FILTER"]
assert vcf_header._indices["FILTER"]["q10a"] is vcf_header.lines[-1]
# Check resulting added header line
assert vcf_header.lines[-1].key == "FILTER"
VALUE = '<ID=q10a,Description="Quality below 10">'
assert vcf_header.lines[-1].value == VALUE
assert len(vcf_header.lines[-1].mapping) == 2
assert vcf_header.lines[-1].mapping["ID"] == "q10a"
assert vcf_header.lines[-1].mapping["Description"] == "Quality below 10"
def test_gt_phase_char_slash():
call = record.Call("sample", vcfpy.OrderedDict([("GT", "0/1")]))
assert call.gt_phase_char == "/"
def test_gt_phase_char_pipe():
call = record.Call("sample", vcfpy.OrderedDict([("GT", "0|1")]))
assert call.gt_phase_char == "|"
def test_is_phased_false():
call = record.Call("sample", vcfpy.OrderedDict([("GT", "0/1")]))
assert call.is_phased is False
def test_is_phased_mixed():
call = record.Call("sample", vcfpy.OrderedDict([("GT", "0/1|2")]))
assert call.is_phased is True
def test_gt_type_filtered_pass():
call = record.Call("sample", vcfpy.OrderedDict([("GT", "1/1"), ("FT", ["PASS"])]))
assert not call.is_filtered()
def test_gt_type_filtered_no_ft():
call = record.Call("sample", vcfpy.OrderedDict([("GT", "1/1")]))
assert not call.is_filtered()
def test_is_variant_hom_alt():
call = record.Call("sample", vcfpy.OrderedDict([("GT", "1/1")]))
assert call.is_variant
def test_is_variant_hom_ref():
call = record.Call("sample", vcfpy.OrderedDict([("GT", "0/0")]))
assert not call.is_variant
def test_gt_type_het():
call = record.Call("sample", vcfpy.OrderedDict([("GT", "0|1")]))
assert call.gt_type == vcfpy.HET
def test_gt_bases_1_1():
call = record.Call("sample", vcfpy.OrderedDict([("GT", "1|1")]))
build_rec([call])
assert call.gt_bases == ("T", "T")
def test_gt_bases_0_2():
call = record.Call("sample", vcfpy.OrderedDict([("GT", "0|2")]))
build_rec([call])
assert call.gt_bases == ("C", "A")
def test_gt_type_hom_alt():
call = record.Call("sample", vcfpy.OrderedDict([("GT", "1/1")]))
assert call.gt_type == vcfpy.HOM_ALT
def test_gt_type_hom_ref():
call = record.Call("sample", vcfpy.OrderedDict([("GT", "0/0")]))
assert call.gt_type == vcfpy.HOM_REF
def test_is_het_het():
call = record.Call("sample", vcfpy.OrderedDict([("GT", "0|1")]))
assert call.is_het
