def _do_water_alignment(seq1,
seq2,
out_fhand,
gap_open=10.0,
gap_extend=0.5,
out_fmt='markx10',
reverse2=False):
seq1_fhand = NamedTemporaryFile()
seq2_fhand = NamedTemporaryFile()
write_seqrecs(seq1, seq1_fhand, 'fasta')
write_seqrecs(seq2, seq2_fhand, 'fasta')
seq1_fhand.flush()
seq2_fhand.flush()
cmd = [
'water', '-asequence', seq1_fhand.name, '-bsequence', seq2_fhand.name,
'-outfile', out_fhand.name, '-gapopen',
str(gap_open), '-gapextend',
str(gap_extend), '-aformat3', out_fmt
]
if reverse2:
cmd.append('-sreverse2')
stdout = open(os.devnull, 'w')
stderr = open(os.devnull, 'w')
subprocess.check_call(cmd, stdout=stdout, stderr=stderr)
def fastaqual_to_fasta(seq_fhand, qual_fhand, out_fhand):
'It converts a fasta and a qual file into a fastq format file'
seqrecords = PairedFastaQualIterator(seq_fhand, qual_fhand)
try:
write_seqrecs(seqrecords, out_fhand.name, 'fastq')
except ValueError, error:
if error_quality_disagree(error):
raise MalformedFile(str(error))
raise
def fastaqual_to_fasta(seq_fhand, qual_fhand, out_fhand):
'It converts a fasta and a qual file into a fastq format file'
seqrecords = PairedFastaQualIterator(seq_fhand, qual_fhand)
try:
write_seqrecs(seqrecords, out_fhand.name, 'fastq')
except ValueError, error:
if error_quality_disagree(error):
raise MalformedFile(str(error))
raise
def _write_seqrecords(seqs, fhand=None, file_format='fastq'):
'It writes a stream of sequences to a file'
if fhand is None:
fhand = NamedTemporaryFile(suffix='.' + file_format.replace('-', '_'))
seqs = _clean_seqrecord_stream(seqs)
try:
write_seqrecs(seqs, fhand, file_format)
except IOError, error:
# The pipe could be already closed
if not 'Broken pipe' in str(error):
raise
def _write_seqrecords(seqs, fhand=None, file_format='fastq'):
'It writes a stream of sequences to a file'
if fhand is None:
fhand = NamedTemporaryFile(suffix='.' + file_format.replace('-', '_'))
seqs = _clean_seqrecord_stream(seqs)
try:
write_seqrecs(seqs, fhand, file_format)
except IOError, error:
# The pipe could be already closed
if not 'Broken pipe' in str(error):
raise
def _do_water_alignment(seq1, seq2, out_fhand, gap_open=10.0, gap_extend=0.5,
out_fmt='markx10', reverse2=False):
seq1_fhand = NamedTemporaryFile()
seq2_fhand = NamedTemporaryFile()
write_seqrecs(seq1, seq1_fhand, 'fasta')
write_seqrecs(seq2, seq2_fhand, 'fasta')
seq1_fhand.flush()
seq2_fhand.flush()
cmd = ['water', '-asequence', seq1_fhand.name, '-bsequence',
seq2_fhand.name, '-outfile', out_fhand.name, '-gapopen',
str(gap_open), '-gapextend', str(gap_extend), '-aformat3', out_fmt]
if reverse2:
cmd.append('-sreverse2')
stdout = open(os.devnull, 'w')
stderr = open(os.devnull, 'w')
subprocess.check_call(cmd, stdout=stdout, stderr=stderr)
def seqio(in_fhands, out_fhand, out_format, copy_if_same_format=True):
'It converts sequence files between formats'
if out_format not in get_setting('SUPPORTED_OUTPUT_FORMATS'):
raise IncompatibleFormatError("This output format is not supported")
in_formats = [remove_multiline(guess_format(fhand)) for fhand in in_fhands]
if len(in_fhands) == 1 and in_formats[0] == out_format:
if copy_if_same_format:
copyfileobj(in_fhands[0], out_fhand)
else:
rel_symlink(in_fhands[0].name, out_fhand.name)
else:
seqs = _read_seqrecords(in_fhands)
try:
write_seqrecs(seqs, out_fhand, out_format)
except ValueError, error:
if error_quality_disagree(error):
raise MalformedFile(str(error))
if 'No suitable quality scores' in str(error):
msg = 'No qualities available to write output file'
raise IncompatibleFormatError(msg)
raise
def seqio(in_fhands, out_fhand, out_format, copy_if_same_format=True):
'It converts sequence files between formats'
if out_format not in get_setting('SUPPORTED_OUTPUT_FORMATS'):
raise IncompatibleFormatError("This output format is not supported")
in_formats = [get_format(fhand) for fhand in in_fhands]
if len(in_fhands) == 1 and in_formats[0] == out_format:
if copy_if_same_format:
copyfileobj(in_fhands[0], out_fhand)
else:
rel_symlink(in_fhands[0].name, out_fhand.name)
else:
seqs = _read_seqrecords(in_fhands)
try:
write_seqrecs(seqs, out_fhand, out_format)
except ValueError, error:
if error_quality_disagree(error):
raise MalformedFile(str(error))
if 'No suitable quality scores' in str(error):
msg = 'No qualities available to write output file'
raise IncompatibleFormatError(msg)
raise