def Write_SeqFile(seq,id) :
masked_seq= Pick70_mask.mask_longAT(seq,"lower")
outfile = '~temp' + string.replace(id[:100],'/','')
out = open(outfile, 'w')
out.write('>'+id+'\n')
out.write(masked_seq +'\n')
out.close()
return outfile
def filter_mask(data_dic, SYMBOL_LIST, TOLERANT, MASK):
new_dic = {}
for key in data_dic.keys():
seq = data_dic[key][4]
os.environ["seq"] = seq
mask = Pick70_mask.mask(seq, SYMBOL_LIST, TOLERANT)
if mask <= MASK:
new_dic[key] = data_dic[key][:]
print "mask:", len(data_dic.keys()), len(new_dic.keys())
return new_dic
if (os.access(genome+".nsq", os.R_OK)) and (os.access(genome+".nin", os.R_OK)) and (os.access(genome+".nhr", os.R_OK)):
pass
else:
fr, fw, fe = os.popen3('formatdb -i $genome -p F')
errormessage = fe.read()
fr.close()
fw.close()
fe.close()
if errormessage!="":
print errormessage
print "program terminated"
sys.exit(1)
maskinput =open('~maskedinput','w')
for id in index:
masked_seq = Pick70_mask.mask_longAT(seq_dic[id],"lower")
maskinput.write('>'+id+'\n')
maskinput.write(masked_seq+'\n')
maskinput.close()
os.environ['input'] = '~maskedinput'
fr, fw, fe = os.popen3('blastall -i "$input" -d $genome -a $cpunum -p blastn -v 50 -b 50 -e 1 -o ~tempblatout -F "m D" -U')
elif (VERSION =="BLAT"):
Pick70_mask.mask_file(tempinput,input+".mask","dust", "lower")
#blat version
fr,fw,fe = os.popen3('blat "$genome" "$input".mask -out=blast -qMask=lower ~tempblatout')
elif (VERSION =="GFCLIENT"):
Pick70_mask.mask_file(tempinput,input+".mask","dust", "lower")
#gfCleint version