def start_blast(cmd, seq, seqString=None, seqDict=None, **kwargs):
"""Run blast and return results."""
p = classutil.FilePopen(cmd, stdin=classutil.PIPE, stdout=classutil.PIPE,
**kwargs)
if seqString is None:
seqString = seq
if seqDict is not None: # write all seqs to nonblocking ifile
for seqID, seq in seqDict.iteritems():
write_fasta(p.stdin, seq)
seqID = None
else: # just write one query sequence
seqID = write_fasta(p.stdin, seqString)
if p.wait(): # blast returned error code
raise OSError('command %s failed' % ' '.join(cmd))
return seqID, p
def repeat_mask(seq, progname='RepeatMasker', opts=()):
'Run RepeatMasker on a sequence, return lowercase-masked string'
## fd, temppath = tempfile.mkstemp()
## ofile = os.fdopen(fd, 'w') # text file
p = classutil.FilePopen([progname] + list(opts), stdin=classutil.PIPE,
stdinFlag=None)
write_fasta(p.stdin, seq, reformatter=lambda x:x.upper()) # save uppercase!
try:
if p.wait():
raise OSError('command %s failed' % ' '.join(p.args[0]))
ifile = file(p._stdin_path + '.masked', 'rU') # text file
try:
for id,title,seq_masked in read_fasta(ifile):
break # JUST READ ONE SEQUENCE
finally:
ifile.close()
finally: # clean up our temp files no matter what happened
p.close() # close temp stdin file
for fpath in glob.glob(p._stdin_path + '.*'):
try:
os.remove(fpath)
except OSError:
pass
return seq_masked # ONLY THE REPEATS ARE IN LOWERCASE NOW
