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 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=-1, 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 os.path.isfile(outfile) and os.path.getsize(outfile) > 0:
if remove_out:
ali = AlignmentExt.from_msa(AlignIO.read(outfile, 'fasta'))
else:
ali = True
else:
ali = False
if remove_out: safe_unlink(outfile)
safe_unlink(msafile)
return ali
def mktmp(cls, alignments, schema='fasta'):
outfile = mktmp_name('.%s' % schema)
if cls.save(alignments, outfile, schema):
return outfile
else:
safe_unlink(outfile)
return None
with open(entry_batch, 'w') as out: out.write(qstr)
try:
cline = BlastDBcmdCommandline(db=db, out=out_file,
entry_batch=entry_batch)
out, err = cline()
if err:
print '\nError in blastdbcmd call:\n%s\n%s\n%s' % (cline, out, err)
return None
#parse results
records = list(SeqIO.parse(out_file, 'fasta'))
return records
except Exception, e:
print e
return None
finally:
safe_unlink(entry_batch)
safe_unlink(out_file)
def _s2s_blast_batch(self, queries, subjects, subject_locs=None, evalue=0.001, command='blastn', **kwargs):
queries_len = len(queries)
subjects_len = len(subjects)
results = [[None for _s in subjects] for _q in queries]
pairs = list(itertools.product(xrange(queries_len), xrange(subjects_len)))
ignore_none_locs = kwargs.pop('ignore_none_locs', False)
shuffle(pairs)
@MultiprocessingBase.data_mapper
@shelf_result
def worker(qs, queries, subjects, subject_locs):
query = queries[qs[0]]
subject = subjects[qs[1]]
if query is None or subject is None: return None
with open(entry_batch, 'w') as out: out.write(qstr)
try:
cline = BlastDBcmdCommandline(db=db, out=out_file,
entry_batch=entry_batch)
out, err = cline()
if err:
print '\nError in blastdbcmd call:\n%s\n%s\n%s' % (cline, out, err)
return None
#parse results
records = list(SeqIO.parse(out_file, 'fasta'))
return records
except Exception, e:
print e
return None
finally:
safe_unlink(entry_batch)
safe_unlink(out_file)
@classmethod
def fetch_results(cls, results, db, what='record'):
'''Fetch records that were found by BLAST search from the database
@param results: an iterable of Bio.Blast.Record.Blast objects (aka BlastRecords)
@param db: name of the BLAST database to get records from
@param what: string, one of "record", "alignment" or "hsp"
@return: list of SeqRecord objects'''
#parse results into queries
queries = []
for record in results:
for alignment in record.alignments:
q = cls.Query(alignment, what)
if not q: continue
