def test_mask_vcf_removes_matching_sites(self, vcf_file, out_file):
"""mask_vcf should remove the given sites from the VCF file"""
mask.mask_vcf([4, 5], vcf_file, out_file)
with open(out_file) as after, open(vcf_file) as before:
assert len(after.readlines()) == len(
before.readlines()) - 1, "Too many lines removed!"
assert "SEQ\t5" not in after.read(), "Correct sites not removed!"
def test_mask_vcf_bails_on_no_chrom(self, tmpdir):
"""mask_vcf should pull a sys.exit() if get_chrom_name returns None"""
bad_vcf = str(tmpdir / "bad.vcf")
with open(bad_vcf, "w") as fh:
fh.write("#")
with pytest.raises(SystemExit) as err:
mask.mask_vcf([], bad_vcf, "")
def test_mask_vcf_cleanup_flag(self, vcf_file, mp_context):
"""mask_vcf should respect the cleanup flag"""
tmp_mask_file = vcf_file + "_maskTemp"
mp_context.setattr(mask, "run_shell_command", lambda *a, **k: None)
mp_context.setattr(mask, "get_chrom_name", lambda f: "SEQ")
mask.mask_vcf([], vcf_file, "", cleanup=True)
assert not os.path.isfile(
tmp_mask_file), "Temporary mask not cleaned up"
mask.mask_vcf([], vcf_file, "", cleanup=False)
assert os.path.isfile(
tmp_mask_file), "Temporary mask cleaned up as expected"
def test_mask_vcf_handles_gz(self, vcf_file, mp_context):
"""mask_vcf should recognize when the in or out files are .gz and call out accordingly"""
def test_shell(call, raise_errors=True):
assert "--gzvcf" in call
assert "| gzip -c" in call
mp_context.setattr(mask, "run_shell_command", test_shell)
mp_context.setattr(mask, "get_chrom_name", lambda f: "SEQ")
# Still worth using the fixture here because writing the mask file uses this path
# Using it for both entries in case god forbid someone starts assuming that path
# valid earlier than it currently is.
in_file = vcf_file + ".gz"
mask.mask_vcf([1, 5], in_file, in_file)
def test_mask_vcf_creates_maskfile(self, vcf_file, mp_context):
"""mask_vcf should create a 1-indexed mask file from the 0-indexed list of sites"""
mask_file = vcf_file + "_maskTemp"
def shell_has_maskfile(call, **kwargs):
assert mask_file in call
mp_context.setattr(mask, "get_chrom_name", lambda f: "SEQ")
mp_context.setattr(mask, "run_shell_command", shell_has_maskfile)
mask.mask_vcf([1, 5], vcf_file, vcf_file, cleanup=False)
assert os.path.isfile(mask_file), "Mask file was not written!"
with open(mask_file) as fh:
assert fh.read() == "SEQ 2\nSEQ 6", "Incorrect mask file written!"