]
while len(spec_to_get) < numspec:
r = random.choice(possible_spec)
spec_to_get.append(r)
possible_spec.remove(r)
print "------"
print "+Getting sequences"
filelist = os.listdir(indir)
numgenes = 0
for each in filelist:
if each.find(".fa") == -1:
continue
infilename = indir + each
inseqs = core.fastaGetDict(infilename)
if all(s in inseqs for s in spec_to_get):
#print "Getting: " + each;
numgenes = numgenes + 1
outfilename = outdir + each
ofile = open(outfilename, "w")
ofile.write("")
ofile.close()
for seq in inseqs:
if seq in spec_to_get:
core.writeSeqOL(outfilename, inseqs[seq], seq)
print "Got " + str(numgenes) + " genes."
print "-----------"
continue
sn = 0
finalseqs = {}
for gid in tmpline:
if sn == 0:
outfilename = outdir + gid + ".fa"
specid = gid[:6]
if gid in main_seq_dict[specid]:
curseq = main_seq_dict[specid][gid]
if remstart == 1 and curseq[0] == "M":
curseq = curseq[1:]
finalseqs[gid] = curseq
sn = sn + 1
if len(finalseqs) == numspec:
core.filePrep(outfilename, "")
for title in finalseqs:
core.writeSeqOL(outfilename, finalseqs[title], ">" + title)
i = i + 1
pstring = "100.0% complete."
sys.stderr.write('\b' * len(pstring) + pstring)
print "\n# " + core.getTime() + " Done!"
print nonorth, "lines skipped."
print "# ======================================================================="
]
while len(spec_to_get) < numspec:
r = random.choice(possible_spec)
spec_to_get.append(r)
possible_spec.remove(r)
print "------"
print "+Getting sequences"
filelist = os.listdir(indir)
numgenes = 0
for each in filelist:
if each.find(".fa") == -1:
continue
infilename = indir + each
inseqs = core.fastaGetDict(infilename)
if all(s in inseqs for s in spec_to_get):
# print "Getting: " + each;
numgenes = numgenes + 1
outfilename = outdir + each
ofile = open(outfilename, "w")
ofile.write("")
ofile.close()
for seq in inseqs:
if seq in spec_to_get:
core.writeSeqOL(outfilename, inseqs[seq], seq)
print "Got " + str(numgenes) + " genes."
print "-----------"
else:
for title in inseqs:
pseqs[title] = pseqs[title] + inseqs[title][x];
inseqs = pseqs;
else:
for title in inseqs:
inseqs[title] = inseqs[title].replace(r1,r2);
writeseqs = {};
for title in inseqs:
if seqkeep != [""]:
if any(s in title for s in seqkeep):
writeseqs[title] = inseqs[title];
else:
writeseqs[title] = inseqs[title];
if len(writeseqs) == len(seqkeep) or seqkeep == [""]:
outfile = open(outfilename, "w");
outfile.write("");
outfile.close();
for title in writeseqs:
# Writes the sequences to the output file.
core.writeSeqOL(outfilename, writeseqs[title], title);
if not os.path.isfile(ins):
pstring = "100.0% complete.";
sys.stderr.write('\b' * len(pstring) + pstring);
print "\n" + core.getTime() + " Done!";
print "==============================================================================================";
pseqs[title] = pseqs[title] + inseqs[title][x]
else:
for title in inseqs:
pseqs[title] = pseqs[title] + inseqs[title][x]
inseqs = pseqs
else:
for title in inseqs:
inseqs[title] = inseqs[title].replace(r1, r2)
writeseqs = {}
for title in inseqs:
if seqkeep != [""]:
if any(s in title for s in seqkeep):
writeseqs[title] = inseqs[title]
else:
writeseqs[title] = inseqs[title]
if len(writeseqs) == len(seqkeep) or seqkeep == [""]:
outfile = open(outfilename, "w")
outfile.write("")
outfile.close()
for title in writeseqs:
# Writes the sequences to the output file.
core.writeSeqOL(outfilename, writeseqs[title], title)
if not os.path.isfile(ins):
pstring = "100.0% complete."
sys.stderr.write('\b' * len(pstring) + pstring)
print "\n" + core.getTime() + " Done!"
print "=============================================================================================="