def seqio(in_fhands, out_fhands, out_format, copy_if_same_format=True):
'It converts sequence files between formats'
in_formats = [guess_format(fhand) for fhand in in_fhands]
if (len(in_formats) == 1 and in_formats[0] == out_format and
hasattr(in_fhands[0], 'name')):
if copy_if_same_format:
copyfileobj(in_fhands[0], out_fhands[0])
else:
rel_symlink(in_fhands[0].name, out_fhands[0].name)
elif len(in_fhands) == 1 and len(out_fhands) == 1:
try:
SeqIO.convert(in_fhands[0], in_formats[0], out_fhands[0],
out_format)
except ValueError as error:
if error_quality_disagree(error):
raise MalformedFile(str(error))
raise
elif (len(in_fhands) == 1 and len(out_fhands) == 2 and
out_format == 'fasta'):
try:
for seq in read_seqrecords([in_fhands[0]]):
SeqIO.write([seq], out_fhands[0], out_format)
SeqIO.write([seq], out_fhands[1], 'qual')
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 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 next(self):
'It returns a packet'
# We are returning lists and not generators (as before) because
# otherwise we would not be able to use multiproces.Pool
packet = []
count = 0
size = self._packet_size
try:
for item in self._iterable:
packet.append(item)
count += 1
if count >= size:
break
except ValueError as error:
if error_quality_disagree(error):
raise MalformedFile(str(error))
raise
if not packet:
raise StopIteration
return packet
def _read_seqitems(fhands):
'it returns an iterator of seq items (tuples of name and chunk)'
seq_iters = []
for fhand in fhands:
file_format = get_format(fhand)
if file_format == 'fasta':
seq_iter = _itemize_fasta(fhand)
elif 'multiline' not in file_format and 'fastq' in file_format:
try:
seq_iter = _itemize_fastq(fhand)
except ValueError as error:
if error_quality_disagree(error):
raise MalformedFile(str(error))
raise
else:
msg = 'Format not supported by the itemizers: ' + file_format
raise NotImplementedError(msg)
seq_iter = assing_kind_to_seqs(SEQITEM, seq_iter, file_format)
seq_iters.append(seq_iter)
return chain.from_iterable(seq_iters)
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
def _read_seqitems(fhands, file_format):
'it returns an iterator of seq items (tuples of name and chunk)'
seq_iters = []
for fhand in fhands:
if file_format == GUESS_FORMAT or file_format is None:
file_format = guess_format(fhand)
else:
file_format = file_format
if file_format == 'fasta':
seq_iter = _itemize_fasta(fhand)
elif 'multiline' not in file_format and 'fastq' in file_format:
try:
seq_iter = _itemize_fastq(fhand)
except ValueError as error:
if error_quality_disagree(error):
raise MalformedFile(str(error))
raise
else:
msg = 'Format not supported by the itemizers: ' + file_format
raise NotImplementedError(msg)
seq_iter = assing_kind_to_seqs(SEQITEM, seq_iter, file_format)
seq_iters.append(seq_iter)
return chain.from_iterable(seq_iters)
try:
for seq in read_seqrecords([in_fhands[0]]):
SeqIO.write([seq], out_fhands[0], out_format)
SeqIO.write([seq], out_fhands[1], 'qual')
except ValueError, error:
if error_quality_disagree(error):
raise MalformedFile(str(error))
raise
elif (len(in_fhands) == 2 and len(out_fhands) == 1 and
in_formats == ['fasta', 'qual']):
seq_records = SeqIO.QualityIO.PairedFastaQualIterator(in_fhands[0],
in_fhands[1])
try:
SeqIO.write(seq_records, out_fhands[0].name, out_format)
except ValueError, error:
if error_quality_disagree(error):
raise MalformedFile(str(error))
raise
elif (len(in_fhands) == 2 and len(out_fhands) == 2 and
in_formats == ['fasta', 'qual'] and out_format == 'fasta'):
if copy_if_same_format:
copyfileobj(in_fhands[0], out_fhands[0])
copyfileobj(in_fhands[1], out_fhands[1])
else:
rel_symlink(in_fhands[0].name, out_fhands[0].name)
rel_symlink(in_fhands[1].name, out_fhands[1].name)
else:
raise RuntimeError('Please fixme, we should not be here')
for out_fhand in out_fhands:
out_fhand.flush()