def convert(in_file, in_format, out_file, out_format, alphabet=None):
"""Convert between two sequence file formats, return number of records.
- in_file - an input handle or filename
- in_format - input file format, lower case string
- out_file - an output handle or filename
- out_format - output file format, lower case string
- alphabet - optional alphabet to assume
NOTE - If you provide an output filename, it will be opened which will
overwrite any existing file without warning. This may happen if even
the conversion is aborted (e.g. an invalid out_format name is given).
For example, going from a filename to a handle:
>>> from Bio import SeqIO
>>> from StringIO import StringIO
>>> handle = StringIO("")
>>> SeqIO.convert("Quality/example.fastq", "fastq", handle, "fasta")
3
>>> print handle.getvalue()
>EAS54_6_R1_2_1_413_324
CCCTTCTTGTCTTCAGCGTTTCTCC
>EAS54_6_R1_2_1_540_792
TTGGCAGGCCAAGGCCGATGGATCA
>EAS54_6_R1_2_1_443_348
GTTGCTTCTGGCGTGGGTGGGGGGG
<BLANKLINE>
"""
if isinstance(in_file, basestring):
#Hack for SFF, will need to make this more general in future
if in_format in _BinaryFormats:
in_handle = open(in_file, "rb")
else:
in_handle = open(in_file, "rU")
in_close = True
else:
in_handle = in_file
in_close = False
#Don't open the output file until we've checked the input is OK?
if isinstance(out_file, basestring):
if out_format in ["sff", "sff_trim"]:
out_handle = open(out_file, "wb")
else:
out_handle = open(out_file, "w")
out_close = True
else:
out_handle = out_file
out_close = False
#This will check the arguments and issue error messages,
#after we have opened the file which is a shame.
from _convert import _handle_convert #Lazy import
count = _handle_convert(in_handle, in_format, out_handle, out_format,
alphabet)
#Must now close any handles we opened
if in_close:
in_handle.close()
if out_close:
out_handle.close()
return count
def convert(in_file, in_format, out_file, out_format, alphabet=None):
"""Convert between two sequence file formats, return number of records.
- in_file - an input handle or filename
- in_format - input file format, lower case string
- out_file - an output handle or filename
- out_format - output file format, lower case string
- alphabet - optional alphabet to assume
NOTE - If you provide an output filename, it will be opened which will
overwrite any existing file without warning. This may happen if even
the conversion is aborted (e.g. an invalid out_format name is given).
For example, going from a filename to a handle:
>>> from Bio import SeqIO
>>> from StringIO import StringIO
>>> handle = StringIO("")
>>> SeqIO.convert("Quality/example.fastq", "fastq", handle, "fasta")
3
>>> print handle.getvalue()
>EAS54_6_R1_2_1_413_324
CCCTTCTTGTCTTCAGCGTTTCTCC
>EAS54_6_R1_2_1_540_792
TTGGCAGGCCAAGGCCGATGGATCA
>EAS54_6_R1_2_1_443_348
GTTGCTTCTGGCGTGGGTGGGGGGG
<BLANKLINE>
"""
if isinstance(in_file, basestring):
# Hack for SFF, will need to make this more general in future
if in_format in _BinaryFormats:
in_handle = open(in_file, "rb")
else:
in_handle = open(in_file, "rU")
in_close = True
else:
in_handle = in_file
in_close = False
# Don't open the output file until we've checked the input is OK?
if isinstance(out_file, basestring):
if out_format in ["sff", "sff_trim"]:
out_handle = open(out_file, "wb")
else:
out_handle = open(out_file, "w")
out_close = True
else:
out_handle = out_file
out_close = False
# This will check the arguments and issue error messages,
# after we have opened the file which is a shame.
from _convert import _handle_convert # Lazy import
count = _handle_convert(in_handle, in_format, out_handle, out_format, alphabet)
# Must now close any handles we opened
if in_close:
in_handle.close()
if out_close:
out_handle.close()
return count