def align(cls, seq_records, outfile=None):
'''Align given sequences
@param seq_records: a list of SeqRecords objects
@param outfile: a filename for the output alignment or None
@return: if the outfile is none, return an AlignmentExt object;
otherwise return True on success. In both cases return None on error.'''
if not outfile:
outfile = mktmp_name('.aln.fasta')
remove_out = True
else: remove_out = False
msafile = mktmp_fasta(seq_records)
args = dict(thread=cpu_count, input=msafile)
if len(seq_records) < 10000:
args['auto'] = True
else:
args['parttree'] = True
args['partsize'] = 1000
ali = None
if run_cline(MafftCommandline(**args), stdout=outfile):
if remove_out:
ali = AlignmentExt.from_msa(AlignIO.read(outfile, 'fasta'))
else: ali = True
if remove_out: safe_unlink(outfile)
safe_unlink(msafile)
return ali
def blast_seq(cls, query, db='nr', evalue=0.001, command='blastn', **kwargs):
'''Perform local blast of a SeqRecord against a database'''
qfile = mktmp_fasta(query)
results = cls.blast(command, query=qfile, db=db,
evalue=evalue, **kwargs)
os.unlink(qfile)
return results
def s2f_blast(cls, query, subject_file, evalue=0.001, command='blastn', **kwargs):
'''Perform local blast of a SeqRecord against a file with sequence(s)'''
qfile = mktmp_fasta(query)
results = cls.blast(command, query=qfile, subject=subject_file,
evalue=evalue, **kwargs)
os.unlink(qfile)
return results
def format_tmp_db(cls, sequences, nucleotide=True):
'''Create a temporary local blast database
@param sequences: SeqRecord object to populate the database with
@param nucleotide: if the sequences are nucleotide (True) or protein (False)
@return: database name that includes includes its path'''
basename = random_text(8)
dbdir = tempfile.mkdtemp('_blastDB')
sfile = mktmp_fasta(sequences)
if run_cline(FormatDBCommandline(input=sfile,
protein='F' if nucleotide else 'T',
name=basename),
cwd=dbdir):
return os.path.join(dbdir, basename)
else: shutil.rmtree(dbdir, ignore_errors=True)
def hmmsearch_recs(hmm, recs, **kwargs):
recfile = mktmp_fasta(recs)
hmm_out = mktmp_name('.hmm.txt')
try:
cline = HMMSearchCommandline(hmmfile=hmm, seqdb=recfile,
o=hmm_out, cpu=cpu_count, seed=0,
**kwargs)
stdout, stderr = cline()
print stdout
if stderr:
sys.stderr.write(stderr)
sys.stderr.flush()
#parse hmmsearch results
with open(hmm_out) as inp:
return list(Hmmer3TextParser(inp))
except Exception, e:
print 'Error while running hmmsearch.'
print e
return None
def hmmsearch_recs(hmm, recs, **kwargs):
recfile = mktmp_fasta(recs)
hmm_out = mktmp_name('.hmm.txt')
try:
cline = HMMSearchCommandline(hmmfile=hmm,
seqdb=recfile,
o=hmm_out,
cpu=cpu_count,
seed=0,
**kwargs)
stdout, stderr = cline()
print stdout
if stderr:
sys.stderr.write(stderr)
sys.stderr.flush()
#parse hmmsearch results
with open(hmm_out) as inp:
return list(Hmmer3TextParser(inp))
except Exception, e:
print 'Error while running hmmsearch.'
print e
return None
def s2s_blast(cls, query, subject, evalue=0.001, command='blastn', **kwargs):
'''Perform local blast of one SeqRecord against the other'''
sfile = mktmp_fasta(subject)
results = cls.s2f_blast(query, sfile, evalue, command, **kwargs)
os.unlink(sfile)
return results
def s2s_blast(cls, query, subject, evalue=0.001, command='blastn', **kwargs):
'''Perform local blast of one SeqRecord against the other'''
sfile = mktmp_fasta(subject)
results = cls.s2f_blast(query, sfile, evalue, command, **kwargs)
os.unlink(sfile)
return results