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 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_wrong_format(self):
fasta = StringIO(
dedent("""
# a comment
# another one
unexpected
>first_sequence
SEQUENCE1
>second_sequence
SEQUENCE2
"""))
reads = list(FastaReader(fasta))
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 parse(self, spec, cmdline_type='back'):
"""
Parse an adapter specification and yield appropriate Adapter classes.
This works like the _parse_no_file() function above, but also supports the
``file:`` notation for reading adapters from an external FASTA
file. Since a file can contain multiple adapters, this
function is a generator.
"""
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]
yield self._parse(record.sequence, cmdline_type, name=name)
else:
yield self._parse(spec, cmdline_type, name=None)
def gather_adapters(back, anywhere, front):
"""
Yield (name, seq, where) tuples from which Adapter instances can be built.
This generator deals with the notation for anchored 5'/3' adapters and also
understands the ``file:`` syntax for reading adapters from an external FASTA
file.
"""
for adapter_list, where in ((back, BACK), (anywhere, ANYWHERE), (front, FRONT)):
for seq in adapter_list:
if seq.startswith('file:'):
# read adapter sequences from a file
path = seq[5:]
with FastaReader(path) as fasta:
for record in fasta:
name = record.name.split(None, 1)[0]
seq, w = parse_adapter(record.sequence, where)
yield (name, seq, w)
else:
name, seq = parse_adapter_name(seq)
seq, w = parse_adapter(seq, where)
yield (name, seq, w)
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'