def test(self):
with FastaReader("tests/data/simple.fasta") as f:
reads = list(f)
assert reads == simple_fasta
fasta = StringIO(">first_sequence\nSEQUENCE1\n>second_sequence\nSEQUENCE2\n")
reads = list(FastaReader(fasta))
assert reads == simple_fasta
def parse_multi(self, back=[], anywhere=[], front=[]):
"""
Parse all three types of commandline options that can be used to
specify adapters. back, anywhere and front are lists of strings,
corresponding to the respective commandline types (-a, -b, -g).
Return a list of appropriate Adapter classes.
"""
adapters = []
for specs, cmdline_type in (back, 'back'), (anywhere,
'anywhere'), (front,
'front'):
for spec in specs:
if spec.startswith('file:'):
# read adapter sequences from a file
with FastaReader(spec[5:]) as fasta:
for record in fasta:
name = record.name.split(None, 1)[0]
adapters.append(
self.parse(record.sequence, name,
cmdline_type))
else:
adapters.append(
self.parse(spec=spec, cmdline_type=cmdline_type))
return adapters
def test_context_manager(self):
filename = "tests/data/simple.fasta"
with open(filename) as f:
assert not f.closed
reads = list(openseq(f))
assert not f.closed
assert f.closed
with FastaReader(filename) as sr:
tmp_sr = sr
assert not sr._file.closed
reads = list(sr)
assert not sr._file.closed
assert tmp_sr._file is None
# Open it a second time
with FastaReader(filename) as sr:
pass
def test_with_comments(self):
fasta = StringIO(dedent(
"""
# a comment
# another one
>first_sequence
SEQUENCE1
>second_sequence
SEQUENCE2
"""))
reads = list(FastaReader(fasta))
assert reads == simple_fasta
def test_wrong_format(self):
with raises(FormatError):
fasta = StringIO(dedent(
"""
# a comment
# another one
unexpected
>first_sequence
SEQUENCE1
>second_sequence
SEQUENCE2
"""))
reads = list(FastaReader(fasta))
def load_from_fasta(self, fasta):
"""
Returns a dict of seq:set(names)
"""
close = False
if isinstance(fasta, str):
fasta = open(fasta, 'rt')
close = True
with FastaReader(fasta) as fasta:
for record in fasta:
name = record.name.split(None, 1)[0]
seq = record.sequence
self.add(name, seq)
if close:
fasta.close()
def test_fastareader_keeplinebreaks(self):
with FastaReader("tests/data/simple.fasta", keep_linebreaks=True) as f:
reads = list(f)
assert reads[0] == simple_fasta[0]
assert reads[1].sequence == 'SEQUEN\nCE2'