def test_write_annotated_record(header_samples, tmpdir_factory):
O = vcfpy.OrderedDict
S = record.Substitution
# open temporary file and setup the Writer with header
path = tmpdir_factory.mktemp("write_annotated_record").join("out.vcf")
header, _ = header_samples
w = writer.Writer.from_path(path, header)
# construct record to write out from scratch
r = record.Record(
"20",
100,
["rs333", "CSN42"],
"C",
[
record.Substitution(record.SNV, "T"),
record.Substitution(record.SNV, "G")
],
50,
["PASS"],
O([("DP", 93), ("AF", [0.3, 0.2]), ("DB", True)]),
["GT", "DP", "GQ", "HQ"],
[
record.Call("NA00001", O(GT="0/1", DP=30, GQ=40, HQ=[1, 2])),
record.Call("NA00002", O(GT="0/2", DP=31, GQ=41, HQ=[3, 4])),
record.Call("NA00003", O(GT="1/2", DP=32, GQ=42, HQ=[5, 6])),
],
)
# write out the record, close file to ensure flushing to disk
w.write_record(r)
w.close()
# compare actual result with expected
RESULT = path.read()
LINE = "20\t100\trs333;CSN42\tC\tT,G\t50\tPASS\tDP=93;AF=0.3,0.2;DB\tGT:DP:GQ:HQ\t0/1:30:40:1,2\t0/2:31:41:3,4\t1/2:32:42:5,6\n"
EXPECTED = MEDIUM_HEADER + LINE
assert EXPECTED == RESULT
def test_write_minimal_record(header_samples, tmpdir_factory):
O = vcfpy.OrderedDict
# open temporary file and setup the Writer with header
path = tmpdir_factory.mktemp("write_header").join("out.vcf")
header, _ = header_samples
w = writer.Writer.from_path(path, header)
# construct record to write out from scratch
r = record.Record(
"20",
100,
[],
"C",
[record.Substitution(record.SNV, "T")],
None,
[],
O(),
["GT"],
[
record.Call("NA00001", O(GT="0/1")),
record.Call("NA00002", O(GT="0/0")),
record.Call("NA00003", O(GT="1/1")),
],
)
# write out the record, close file to ensure flushing to disk
w.write_record(r)
w.close()
# compare actual result with expected
RESULT = path.read()
LINE = "20\t100\t.\tC\tT\t.\t.\t.\tGT\t0/1\t0/0\t1/1\n"
EXPECTED = MEDIUM_HEADER + LINE
assert EXPECTED == RESULT
def test_write_record_with_escaping(header_samples, tmpdir_factory):
O = vcfpy.OrderedDict
S = record.Substitution
# open temporary file and setup the Writer with header
path = tmpdir_factory.mktemp("write_header").join("out.vcf")
header, _ = header_samples
w = writer.Writer.from_path(path, header)
# construct record to write out from scratch
r = record.Record(
"20",
100,
[],
"C",
[record.Substitution(record.SNV, "T")],
None,
[],
O([("ANNO", ["Here,are%some chars", "%25"])]),
["GT", "FT"],
[
record.Call("NA00001", O(GT="0/1", FT=["%25", "FOO"])),
record.Call("NA00002", O(GT="0/0", FT=[])),
record.Call("NA00003", O(GT="1/1", FT=[])),
],
)
# write out the record, close file to ensure flushing to disk
w.write_record(r)
w.close()
# compare actual result with expected
RESULT = path.read()
LINE = ("20\t100\t.\tC\tT\t.\t.\tANNO=Here%2Care%25some chars,"
"%2525\tGT:FT\t0/1:%2525;FOO\t0/0:.\t1/1:.\n")
EXPECTED = MEDIUM_HEADER + LINE
assert EXPECTED == RESULT
def test_write_minimal_record_writer_from_stream_path(header_samples, tmpdir_factory):
O = vcfpy.OrderedDict
# open temporary file and setup the Writer with header
path = tmpdir_factory.mktemp("write_header").join("out.vcf.gz")
header, _ = header_samples
with open(str(path), "wb") as f:
w = writer.Writer.from_stream(f, header, path=str(path))
# construct record to write out from scratch
r = record.Record(
"20",
100,
[],
"C",
[record.Substitution(record.SNV, "T")],
None,
[],
O(),
["GT"],
[
record.Call("NA00001", O(GT="0/1")),
record.Call("NA00002", O(GT="0/0")),
record.Call("NA00003", O(GT="1/1")),
],
)
# write out the record, close file to ensure flushing to disk
w.write_record(r)
w.close()
# check the resulting record
LINE = "20\t100\t.\tC\tT\t.\t.\t.\tGT\t0/1\t0/0\t1/1\n"
check_file(path, LINE)
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_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