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 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 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)
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 hmmbuild(alignment, outfile, name=None, **kwargs):
unlink_file = False
if isinstance(alignment, str): msafile = alignment
elif isinstance(alignment, MultipleSeqAlignment):
msafile = AlignmentUtils.mktmp(alignment)
if not msafile: return False
unlink_file = True
else:
print 'Alignment must be either a filename or an instance of MultipleSeqAlignment'
return False
if not name: name = FilenameParser.strip_ext(os.path.basename(outfile))
ret = run_cline(HMMBuildCommandline(input=msafile,
out=outfile,
n=name,
cpu=cpu_count,
seed=0,
**kwargs),
_msg='Unable to build HMM profile')
if unlink_file: os.unlink(msafile)
return ret
def hmmbuild(alignment, outfile, name=None, **kwargs):
msafile = AlignmentUtils.mktmp(alignment)
if not name: name = FilenameParser.strip_ext(outfile)
return run_cline(HMMBuildCommandline(input=msafile, out=outfile,
n=name, cpu=cpu_count, seed=0, **kwargs),
_msg = 'Unable to build HMM profile')