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
queries.append(q)
if not queries: return None
#fetch records from database
out_file = mktmp_name('.fasta')
entry_batch = mktmp_name('.qry')
qstr = '\n'.join(str(q) for q in queries)
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
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
def fetch_queries(cls, queries, db):
out_file = mktmp_name('.fasta')
entry_batch = mktmp_name('.qry')
qstr = '\n'.join(str(q) for q in queries)
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
def build_fast_tree(cls, alignment, outfile):
'''Build an approximate-ML tree with fasttree.
@param outfile: output filename
@param alignment: alignment object or a filename of an alignment in fasta format'''
print 'Building an approximate-ML tree with fasttree.'
if isinstance(alignment, MultipleSeqAlignment):
alnfile = mktmp_name('.aln.fasta')
cls.save(alignment, alnfile)
elif isinstance(alignment, basestring):
alnfile = alignment
else: raise TypeError('Unsupported type of alignment argument: %s' % type(alignment))
args = dict(input=alnfile, out=outfile, pseudo=1)
if num_fasta_records(alnfile) >= 10000:
args['fastest'] = True
args['boot'] = 100
if not run_cline(FastTreeCommandline(**args), name=cls.strip_ext(outfile)):
return False
print 'Done\n'
return 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 blast(cls, command, **kwargs):
'''Generic wrapper for blast commandline programs'''
results_file = kwargs.pop('save_results_to', None)
parse_results = kwargs.pop('parse_results', False)
if results_file: bout = results_file
else:
parse_results = True
bout = mktmp_name('.xml')
try:
cline = cls._clines[command]
cmd = cline(outfmt=5, out=bout, **kwargs)
out, err = cmd()
if err:
print ('\nError message from %s:\n%s\n%s\n%s'
% (command, cmd, out, err))
if parse_results:
with open(bout) as inp:
results = list(NCBIXML.parse(inp))
return results
except Exception, e:
print '\nError while performing local %s:\n%s' % (command, estr(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 blast(cls, command, **kwargs):
'''Generic wrapper for blast commandline programs'''
results_file = kwargs.pop('save_results_to', None)
parse_results = kwargs.pop('parse_results', False)
if results_file: bout = results_file
else:
parse_results = True
bout = mktmp_name('.xml')
try:
cline = cls._clines[command]
cmd = cline(outfmt=5, out=bout, **kwargs)
out, err = cmd()
if err:
print ('\nError while performing local %s:\n%s\n%s\n%s'
% (command, cmd, out, err))
return None
if parse_results:
with open(bout) as inp:
results = list(NCBIXML.parse(inp))
return results
except Exception, e:
print '\nError while performing local %s:' % command
print e
return None
def mktmp(cls, alignments, schema='fasta'):
outfile = mktmp_name('.%s' % schema)
cls.save(alignments, outfile, schema)
