def BLAST(seq, seqs, evalue=10):
lib_dir = temp('blast-' + str(id(seqs)))
res_file = temp('blast-' + str(id(seqs)), str(id(seq)))
dbtype = 'nucl' if isinstance(seqs[0], NASeq) else 'prot'
seqtype = 'nucl' if isinstance(seq, NASeq) else 'prot'
if not op.exists(op.join(lib_dir, seqs.get_file('fa') + '.phr')) \
and not op.exists(op.join(lib_dir, seqs.get_file('fa') + '.nhr')):
c = [[app('BLAST','makeblastdb'),'-in',seqs.get_file('fa'),'-out',lib_dir,'-parse_seqids','-dbtype',dbtype]]
else:
c = []
if dbtype == 'nucl' and seqtype == 'nucl':
blprog = 'blastn' # TBLASTX
elif dbtype == 'nucl' and seqtype == 'prot':
blprog = 'blastx'
elif dbtype == 'prot' and seqtype == 'nucl':
blprog = 'tblastn'
elif dbtype == 'prot' and seqtype == 'prot':
blprog = 'blastp' # PSIBLAST, PHIBLAST
c += [[app('BLAST',blprog),'-db',lib_dir,'-query',seq.get_file('fa'),'-outfmt','"6 sseqid"','-out',res_file+'-out','-evalue',str(evalue)]]
c += [[app('BLAST','blastdbcmd'),'-db',lib_dir,'-entry_batch',res_file+'-out','-out',res_file+'.fa']]
run(c)
return SeqList(open(res_file + '.fa', 'r'))
def get_file(self, frmt="fa", fdir=None):
"""
Write my sequences to disk in a file *seqs-######* in the
directory *fdir* making sure they're in a format specified by the list "frmt"
and return a FileSeqList.
Useful for quickly getting a file in the correct format to
pass onto an external program.
Parameters:
frmt (str): a file type (e.g. FASTA, FASTQ, etc)
(default: FASTA)
fdir (str): a directory to save the file in
(default: the temp directory)
Returns:
the file name of the saved sequences
"""
from ochre.External.External import temp
if isinstance(frmt, str):
frmt = (frmt,)
fname = "seqs-" + str(id(self)) + "." + frmt[0]
if fdir is None:
file_name = temp(fname)
else:
file_name = os.path.join(fdir, fname)
self.write(file_name, frmt[0])
return file_name
