def blast(db_seq,
qry_seq,
seqtype='nucl',
blast_out=None,
blast_outfmt=None,
ncpu=4):
if seqtype == 'nucl':
blast_app = 'blastn'
elif seqtype == 'prot':
blast_app = 'blastp'
else:
raise ValueError(
'Unknown molecule type "{}" for blast'.format(seqtype))
if blast_out is None:
blast_out = qry_seq + '.blastout'
if blast_outfmt is None:
blast_outfmt = '6'
cmd = 'makeblastdb -in {} -dbtype {}'.format(db_seq, seqtype)
run_cmd(cmd, logger=logger)
cmd = '{} -query {} -db {} -out {} -outfmt {} -num_threads {}'.format(
blast_app, qry_seq, db_seq, blast_out, blast_outfmt, ncpu)
# cmd += " " + blast_opts
run_cmd(cmd, logger=logger)
return blast_out
def hmmscan(inSeq, hmmdb='rexdb.hmm', hmmout=None, ncpu=4):
if hmmout is None:
hmmout = prefix + '.domtbl'
cmd = 'hmmscan --notextw -E 0.01 --domE 0.01 --noali --cpu {} --domtblout {} {} {} > /dev/null'.format(
ncpu, hmmout, hmmdb, inSeq)
run_cmd(cmd, logger=logger)
return hmmout
def main(inSeq, domains, outSeq=sys.stdout, tmpdir='/tmp'):
d_domain = {domain: [] for domain in domains}
uid = uuid.uuid1()
# intersect
for rc in SeqIO.parse(inSeq, 'fasta'):
domain = re.compile(r'gene=([^;\s]+)').search(
rc.description).groups()[0]
if domain in d_domain:
raw_id = '#'.join(rc.id.split('#')[:-1])
d_domain[domain] += [raw_id]
i = 0
for domain, rawids in d_domain.iteritems():
i += 1
if i == 1:
intersect = set(rawids)
continue
intersect = intersect & set(rawids)
print >> sys.stderr, '{} sequences contain {} domains'.format(
len(intersect), domains)
# open files
d_file = {}
files = []
for i, domain in enumerate(domains):
outfile = '{}/{}.{}.fa'.format(tmpdir, uid, i)
fout = open(outfile, 'w')
d_file[domain] = fout
files += [outfile]
# write
for rc in SeqIO.parse(inSeq, 'fasta'):
domain = re.compile(r'gene=([^;\s]+)').search(
rc.description).groups()[0]
if not domain in d_file:
continue
raw_id = '#'.join(rc.id.split('#')[:-1])
if raw_id not in intersect:
continue
fout = d_file[domain]
rc.id = raw_id
SeqIO.write(rc, fout, 'fasta')
# close files
for fout in d_file.values():
fout.close()
# align
alnfiles = []
for seqfile in files:
alnfile = seqfile + '.aln'
cmd = 'mafft --auto {} > {}'.format(seqfile, alnfile)
# os.system(cmd)
run_cmd(cmd, log=True)
alnfiles += [alnfile]
# concatenate
catAln(alnfiles, outSeq)
def update_hmmer(self, db):
from small_tools import backup_file
bk_db, db = backup_file(db)
for suffix in ['.h3f', '.h3i', '.h3m', '.h3p']:
backup_file(bk_db + suffix)
cmd = 'hmmconvert {} > {}'.format(bk_db, db)
out, err, status0 = run_cmd(cmd, logger=logger)
cmd = 'hmmpress {}'.format(db)
out, err, status1 = run_cmd(cmd, logger=logger)
if status0 + status1 == 0:
logger.info('HMM converted. it will continue')
else:
logger.error('HMM failed to convert. exit')
sys.exit(1)
def check_presence(self, program):
cmd = 'which {}'.format(program)
out, err, status = run_cmd(cmd)
if status == 0:
return True
else:
return False
def pipeline(self):
for clade, records in self.BinSeqsByClade().iteritems():
clade0 = clade.replace('/', '__').replace(':', '--')
cladeSeqs = '{}/{}.fa'.format(self.tmpdir, clade0)
with open(cladeSeqs, 'w') as f:
for rc in records:
SeqIO.write(rc, f, 'fasta')
alnSeqs = cladeSeqs + '.aln'
cmd = 'mafft --auto {} > {} 2> /dev/null'.format(
cladeSeqs, alnSeqs)
# cmd = 'mafft --auto {} 2> /dev/null| prepareAlign | mafft --auto - > {} 2> /dev/null'.format(cladeSeqs, alnSeqs)
run_cmd(cmd)
alnHMM = cladeSeqs + '.hmm'
cmd = 'hmmbuild -n {} {} {} > /dev/null'.format(
clade, alnHMM, alnSeqs)
run_cmd(cmd)
def check_blast_version(self, program):
cmd = '{} -version'.format(program)
out, err, status = run_cmd(cmd)
version = re.compile(r'blast\S* ([\d\.\+]+)').search(out).groups()[0]
return version
def check_hmmer_verion(self, program):
cmd = '{} -h'.format(program)
out, err, status = run_cmd(cmd)
version = re.compile(r'HMMER (\S+)').search(out).groups()[0]
return version
