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 Trimmomatic(seqs, headcrop=None, leading=None, trailing=None, slidingwin=None, minlen=None, illuminaclip=None):
if isinstance(seqs, PairedFileSeqList):
tend = 'PE'
f1, f2 = seqs._f1.get_file('fq'), seqs._f2.get_file('fq')
flist = [f1, f2, \
f1 + '.trim.prd.fq', f1 + '.trim.uprd.fq', \
f2 + '.trim.prd.fq', f2 + '.trim.uprd.fq']
elif isinstance(seqs, SeqList):
tend = 'SE'
f1 = seqs.get_file('fq')
flist = [f1, f1 + '.trim.fq']
else:
raise TypeError('Need a PairedFileSeqList, FileSeqList or SeqList.')
qual = '-phred33' # TODO: determine this from the data
c = [app('JAVA', 'java'), '-classpath', \
app('TRIMMOMATIC', 'trimmomatic-0.22.jar'), \
'org.usadellab.trimmomatic.Trimmomatic' + tend, qual]
c += flist
if illuminaclip is not None:
c += ['ILLUMINACLIP:' + str(illuminaclip.get_file('fa')) + \
':2:40:15'] # sane defaults?
if leading is not None:
c += ['LEADING:' + str(leading)]
if trailing is not None:
c += ['TRAILING:' + str(trailing)]
if slidingwin is not None:
c += ['SLIDINGWINDOW:' + str(slidingwin[0]) + ':' + \
str(slidingwin[1])]
if headcrop is not None:
c += ['HEADCROP:' + str(headcrop)]
if minlen is not None:
c += ['MINLEN:' + str(minlen)]
run([c])
if tend == 'PE':
return PairedFileSeqList(f1 + '.trim.prd.fq',\
f2 + '.trim.prd.fq')
else:
return SeqList(f1 + '.trim.fq')