def _do_search(self, seq, seqdb, args):
seq_file = tempfile.NamedTemporaryFile()
seqdb_file = tempfile.NamedTemporaryFile()
out_file = tempfile.NamedTemporaryFile()
hmm_file = tempfile.NamedTemporaryFile()
SeqIO.write(seq, seq_file, 'fasta')
SeqIO.write(seqdb, seqdb_file, 'fasta')
seq_file.flush()
seqdb_file.flush()
p = Popen(['jackhmmer', '--qformat', 'fasta', '--tformat', 'fasta',
'--chkhmm', hmm_file.name, '--domtblout', out_file.name,] + args +
[seq_file.name, seqdb_file.name,],
stdout=PIPE, stdin=PIPE, stderr=PIPE)
out = p.communicate()
self.matches += matchfile.load(out_file, seq, self.seqdb)
#load the hmms
try:
i = 1
while True:
f = "{}-{}.hmm".format(hmm_file.name, i)
self.hmms.append(hmmfile.read(f)[0])
os.remove(f)
i += 1
except IOError:
pass
out_file.close()
seq_file.close()
seqdb_file.close()
def _do_search(self, hmm, target, args):
#bail if there isn't any sequence to search
if not len(target):
return []
#write the HMM to a temporary file
hmm_file = tempfile.NamedTemporaryFile()
target_file = tempfile.NamedTemporaryFile()
out_file = tempfile.NamedTemporaryFile()
hmmfile.write(hmm, hmm_file)
hmm_file.flush()
SeqIO.write([target,], target_file, 'fasta')
target_file.flush()
p = Popen(['hmmsearch',] + args + ['--tformat', 'fasta',
'--domtblout', out_file.name, hmm_file.name, target_file.name,]
, stdout=PIPE, stdin=PIPE, stderr=PIPE, preexec_fn=set_limits)
out = p.communicate()
if p.returncode != 0:
if p.returncode == -6 and "alloc" in out[1].lower():
seq = target.seq
split = len(seq) / 2
target.seq = seq[0:split]
m_left = self._do_search(hmm, target, args)
target.seq = seq[split:]
m_right = self._do_search(hmm, target, args)
for m in m_right:
m.offset(split)
target.seq = seq
return m_left + m_right
raise RuntimeError('hmmsearch error ({}) : {}'.format(
p.returncode, out[1]))
else:
matches = matchfile.load(out_file, hmm, self.targets)
out_file.close()
hmm_file.close()
target_file.close()
return matches