def test_interleaved_no_paired_output():
with temporary_path("temp-paired.1.fastq") as p1:
with temporary_path("temp-paired.2.fastq") as p2:
params = '-a XX --interleaved'.split()
with redirect_stderr():
params += [ '-o', p1, '-p1', p2, 'paired.1.fastq', 'paired.2.fastq']
atropos.main(params)
def test_no_trimming_legacy():
# make sure that this doesn't divide by zero
atropos.main([
'-a', 'XXXXX', '-o', '/dev/null', '-p', '/dev/null', '-pe1',
datapath('paired.1.fastq'), '-pe2',
datapath('paired.2.fastq')
])
def test_E3M():
'''Read the E3M dataset'''
# not really colorspace, but a fasta/qual file pair
main([
'-o', '/dev/null', '-se',
datapath("E3M.fasta"), '-sq',
datapath("E3M.qual")
])
def test_interleaved_no_paired_output():
with temporary_path("temp-paired.1.fastq") as p1:
with temporary_path("temp-paired.2.fastq") as p2:
params = '-a XX --interleaved'.split()
with raises(SystemExit), redirect_stderr():
params += [
'-o', p1, '-p1', p2, 'paired.1.fastq', 'paired.2.fastq'
]
atropos.main(params)
def test_second_too_short():
with temporary_path("truncated.2.fastq") as trunc2:
# Create a truncated file in which the last read is missing
with open(datapath('paired.2.fastq')) as f:
lines = f.readlines()
lines = lines[:-4]
with open(trunc2, 'w') as f:
f.writelines(lines)
with redirect_stderr():
atropos.main('-a XX --paired-output out.fastq'.split() + [datapath('paired.1.fastq'), trunc2])
def test_second_too_short():
with temporary_path("truncated.2.fastq") as trunc2:
# Create a truncated file in which the last read is missing
with open(datapath('paired.2.fastq')) as f:
lines = f.readlines()
lines = lines[:-4]
with open(trunc2, 'w') as f:
f.writelines(lines)
with raises(SystemExit), redirect_stderr():
atropos.main('-a XX --paired-output out.fastq'.split() +
[datapath('paired.1.fastq'), trunc2])
def test_quiet_is_quiet():
captured_standard_output = StringIO()
captured_standard_error = StringIO()
try:
old_stdout = sys.stdout
old_stderr = sys.stderr
sys.stdout = captured_standard_output
sys.stderr = captured_standard_error
atropos.main(['-o', '/dev/null', '--quiet', '-a', 'XXXX', '-se', datapath('illumina.fastq.gz')])
finally:
sys.stdout = old_stdout
sys.stderr = old_stderr
assert captured_standard_output.getvalue() == ''
assert captured_standard_error.getvalue() == ''
def run(params,
expected,
inpath,
inpath2=None,
qualfile=None,
interleaved_input=False,
interleaved_output=False):
if type(params) is str:
params = params.split()
with temporary_path(expected) as tmp_fastaq:
if interleaved_input:
params += ['-l', inpath]
elif inpath2:
params += ['-pe1', datapath(inpath)]
params += ['-pe2', datapath(inpath2)]
else:
params += ['-se', datapath(inpath)]
if qualfile:
params += ['-sq', datapath(qualfile)]
if interleaved_output:
params += ['-L', tmp_fastaq]
else:
params += ['-o', tmp_fastaq] # TODO not parallelizable
print(params)
result = atropos.main(params)
assert isinstance(result, tuple)
assert len(result) == 3
# TODO redirect standard output
assert files_equal(cutpath(expected), tmp_fastaq)
def run_paired(params,
in1,
in2,
expected1,
expected2,
aligners=('adapter', ),
callback=None,
assert_files_equal=True,
error_on_rc=True):
if type(params) is str:
params = params.split()
for aligner in aligners:
with temporary_path('tmp1-' + expected1.format(aligner=aligner)) as p1:
with temporary_path('tmp2-' +
expected2.format(aligner=aligner)) as p2:
p = params.copy()
p += ['--aligner', aligner, '-o', p1, '-p', p2]
p += [
'-pe1',
datapath(in1.format(aligner=aligner)), '-pe2',
datapath(in2.format(aligner=aligner))
]
result = atropos.main(p)
assert isinstance(result, tuple)
assert len(result) == 3
if error_on_rc:
assert result[0] == 0, "Return code {} != 0".format(
result[0])
if assert_files_equal:
assert files_equal(
cutpath(expected1.format(aligner=aligner)), p1)
assert files_equal(
cutpath(expected2.format(aligner=aligner)), p2)
if callback:
callback(aligner, p1, p2, result)
def run_interleaved(params, inpath, expected, aligners=('adapter',)):
if type(params) is str:
params = params.split()
for aligner in aligners:
with temporary_path(expected.format(aligner=aligner)) as tmp:
p = params.copy()
p += ['--aligner', aligner, '-l', datapath(inpath.format(aligner=aligner)), '-L', tmp]
result = atropos.main(p)
assert isinstance(result, tuple)
assert len(result) == 3
assert files_equal(cutpath(expected.format(aligner=aligner)), tmp)
def test_demultiplex():
multiout = os.path.join(os.path.dirname(__file__), 'data', 'tmp-demulti.{name}.fasta')
params = ['-a', 'first=AATTTCAGGAATT', '-a', 'second=GTTCTCTAGTTCT', '-o', multiout, '-se', datapath('twoadapters.fasta')]
result = atropos.main(params)
assert isinstance(result, tuple)
assert len(result) == 3
assert result[0] == 0
assert files_equal(cutpath('twoadapters.first.fasta'), multiout.format(name='first'))
assert files_equal(cutpath('twoadapters.second.fasta'), multiout.format(name='second'))
assert files_equal(cutpath('twoadapters.unknown.fasta'), multiout.format(name='unknown'))
os.remove(multiout.format(name='first'))
os.remove(multiout.format(name='second'))
os.remove(multiout.format(name='unknown'))
def run_interleaved(params, inpath, expected, aligners=('adapter', )):
if type(params) is str:
params = params.split()
for aligner in aligners:
with temporary_path(expected.format(aligner=aligner)) as tmp:
p = params.copy()
p += [
'--aligner', aligner, '-l',
datapath(inpath.format(aligner=aligner)), '-L', tmp
]
result = atropos.main(p)
assert isinstance(result, tuple)
assert len(result) == 3
assert files_equal(cutpath(expected.format(aligner=aligner)), tmp)
def test_unmatched_read_names():
with temporary_path("swapped.1.fastq") as swapped:
try:
# Create a file in which reads 2 and are swapped
with open(datapath('paired.1.fastq')) as f:
lines = f.readlines()
lines = lines[0:4] + lines[8:12] + lines[4:8] + lines[12:]
with open(swapped, 'w') as f:
f.writelines(lines)
with redirect_stderr():
result = atropos.main(
'-a XX -o out1.fastq -p out2.fastq'.split() +
['-pe1', swapped, '-pe2', datapath('paired.2.fastq')])
assert isinstance(result, tuple)
assert len(result) == 3
assert result[0] != 0
finally:
os.remove('out1.fastq')
os.remove('out2.fastq')
def run_paired(params, in1, in2, expected1, expected2, aligners=('adapter',),
callback=None, assert_files_equal=True):
if type(params) is str:
params = params.split()
for aligner in aligners:
with temporary_path('tmp1-' + expected1.format(aligner=aligner)) as p1:
with temporary_path('tmp2-' + expected2.format(aligner=aligner)) as p2:
p = params.copy()
p += ['--aligner', aligner, '-o', p1, '-p', p2]
p += ['-pe1', datapath(in1.format(aligner=aligner)),
'-pe2', datapath(in2.format(aligner=aligner))]
result = atropos.main(p)
assert isinstance(result, tuple)
assert len(result) == 3
if assert_files_equal:
assert files_equal(cutpath(expected1.format(aligner=aligner)), p1)
assert files_equal(cutpath(expected2.format(aligner=aligner)), p2)
if callback:
callback(aligner, p1, p2, result)
def test_unmatched_read_names():
with temporary_path("swapped.1.fastq") as swapped:
try:
# Create a file in which reads 2 and are swapped
with open(datapath('paired.1.fastq')) as f:
lines = f.readlines()
lines = lines[0:4] + lines[8:12] + lines[4:8] + lines[12:]
with open(swapped, 'w') as f:
f.writelines(lines)
with redirect_stderr():
result = atropos.main(
'-a XX -o out1.fastq -p out2.fastq'.split() +
['-pe1', swapped, '-pe2',
datapath('paired.2.fastq')])
assert isinstance(result, tuple)
assert len(result) == 3
assert result[0] != 0
finally:
os.remove('out1.fastq')
os.remove('out2.fastq')
def run(params, expected, inpath, inpath2=None, qualfile=None, interleaved_input=False, interleaved_output=False):
if type(params) is str:
params = params.split()
with temporary_path(expected) as tmp_fastaq:
if interleaved_input:
params += ['-l', inpath]
elif inpath2:
params += ['-pe1', datapath(inpath)]
params += ['-pe2', datapath(inpath2)]
else:
params += ['-se', datapath(inpath)]
if qualfile:
params += ['-sq', datapath(qualfile)]
if interleaved_output:
params += ['-L', tmp_fastaq]
else:
params += ['-o', tmp_fastaq] # TODO not parallelizable
print(params)
result = atropos.main(params)
assert isinstance(result, tuple)
assert len(result) == 3
# TODO redirect standard output
assert files_equal(cutpath(expected), tmp_fastaq)
def test_qualfile_only():
with redirect_stderr():
atropos.main(['-sq', datapath('E3M.qual')])
def test_no_trimming_legacy():
# make sure that this doesn't divide by zero
atropos.main(['-a', 'XXXXX', '-o', '/dev/null', '-p', '/dev/null', '-pe1', datapath('paired.1.fastq'), '-pe2', datapath('paired.2.fastq')])
def test_missing_file():
with raises(SystemExit), redirect_stderr():
atropos.main([
'-a', 'XX', '--paired-output', 'out.fastq',
datapath('paired.1.fastq')
])
def test_missing_file():
with redirect_stderr():
atropos.main(['-a', 'XX', '--paired-output', 'out.fastq', datapath('paired.1.fastq')])
def test_two_fastqs():
with redirect_stderr():
atropos.main(['-pe1', datapath('paired.1.fastq'), '-pe2', datapath('paired.2.fastq')])
def test_no_args():
with redirect_stderr():
result = atropos.main([])
assert isinstance(result, tuple)
assert len(result) == 3
assert result[0] != 0
def test_E3M():
'''Read the E3M dataset'''
# not really colorspace, but a fasta/qual file pair
main(['-o', '/dev/null', '-se', datapath("E3M.fasta"), '-sq', datapath("E3M.qual")])
