def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
infile = ""
gramfile = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
infile = argv[i]
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-o", "--o", "-outfile"]:
(outfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-gram", "--gram"]:
(gramfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
infile = argv[i]
i += 1
if myfunc.checkfile(infile) != 0:
return 1
if myfunc.checkfile(gramfile) != 0:
return 1
grampairlist = myfunc.ReadPairList(gramfile)
gramMapDict = {}
for tup in grampairlist:
gramMapDict[tup[0]] = tup[1]
gi2taxidpairlist = myfunc.ReadPairList(infile)
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
for tup in gi2taxidpairlist:
try:
fpout.write("%s\t%s\t%s\n" % (tup[0], tup[1], gramMapDict[tup[1]]))
except KeyError:
fpout.write("%s\t%s\t%s\n" % (tup[0], tup[1], "NA"))
myfunc.myclose(fpout)
def main():#{{{
numArgv=len(sys.argv)
if numArgv < 2:
PrintHelp()
return 1
argv = sys.argv
topofile = ""
sigpepfile = ""
outfile = ""
isQuiet = False
isDeleteSeqWithSignalPeptide = False
i = 1
isNonOptionArg=False
while i < numArgv:
if isNonOptionArg == True:
isNonOptionArg = False
i += 1
elif sys.argv[i] == "--":
isNonOptionArg=True
i += 1
elif sys.argv[i][0] == "-":
if sys.argv[i] in [ "-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in [ '-o', '--o', "-outfile", "--outfile"]:
outfile, i = myfunc.my_getopt_str(argv, i)
elif argv[i] in ['-topo', '--topo']:
topofile, i = myfunc.my_getopt_str(argv, i)
elif argv[i] in ['-sig', '--sig']:
sigpepfile, i = myfunc.my_getopt_str(argv, i)
elif argv[i] == "-q":
isQuiet=True; i += 1
elif argv[i] in ["-deleteseq", "--deleteseq"]:
isDeleteSeqWithSignalPeptide=True; i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
if myfunc.checkfile(topofile, 'topofile') != 0:
return 1
if myfunc.checkfile(sigpepfile, 'sigpepfile') != 0:
return 1
sigpepDict = lcmp.ReadSignalPDict(sigpepfile)
FilterSignalPeptide(topofile, sigpepDict, outfile, isDeleteSeqWithSignalPeptide)
return 0
def main(g_params):#{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
infile = ""
mapfile = ""
i = 1
isNonOptionArg=False
while i < numArgv:
if isNonOptionArg == True:
infile = argv[i]
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-o", "--o", "-outfile"]:
(outfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-map", "--map", "-mapfile"]:
(mapfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
infile = argv[i]
i += 1
if myfunc.checkfile(infile) != 0:
return 1
if myfunc.checkfile(mapfile) != 0:
return 1
clanid2pfamidDict = myfunc.ReadFam2SeqidMap(mapfile)
pfamPercentTMDict = ReadPercentTM(infile)
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
GetPercentTMOfClan(pfamPercentTMDict, clanid2pfamidDict, fpout)
myfunc.myclose(fpout)
def main(g_params): #{{{#{{{
# Check argv
numArgv = len(sys.argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
infile = ""
BLOCK_SIZE = 100000
isPrintID = False
isJustPrintSum = False
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
isNonOptionArg = False
infile = sys.argv[i]
i = i + 1
elif sys.argv[i] == "--":
isNonOptionArg = True
i = i + 1
elif sys.argv[i][0] == "-":
if sys.argv[i] in ["-h", "--help"]:
PrintHelp()
return 0
elif sys.argv[i] in ["-i", "--i", "-printid", "--printid"]:
isPrintID = True
i += 1
elif sys.argv[i] in ["-just-print-sum", "--just-print-sum"]:
isJustPrintSum = True
i += 1
elif sys.argv[i] in ["-o", "--o", "-outfile", "--outfile"]:
outfile, i = myfunc.my_getopt_str(sys.argv, i)
elif sys.argv[i] in ["-bs", "--bs", "-block-size", "--block-size"]:
BLOCK_SIZE, i = myfunc.my_getopt_int(sys.argv, i)
if BLOCK_SIZE < 0:
print("Error! BLOCK_SIZE should >0", file=sys.stderr)
return 1
else:
print(("Error! Wrong argument:%s" % sys.argv[i]),
file=sys.stderr)
return 1
else:
infile = sys.argv[i]
i += 1
if myfunc.checkfile(infile) != 0:
return 1
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
status = Getseqlen(infile, isPrintID, isJustPrintSum, BLOCK_SIZE, fpout)
myfunc.myclose(fpout)
return status
def main(g_params):#{{{#{{{
# Check argv
numArgv=len(sys.argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
infile = ""
BLOCK_SIZE = 100000
isPrintID = False
isJustPrintSum = False
i = 1
isNonOptionArg=False
while i < numArgv:
if isNonOptionArg == True:
isNonOptionArg=False
infile = sys.argv[i]
i = i + 1
elif sys.argv[i] == "--":
isNonOptionArg=True
i = i + 1
elif sys.argv[i][0] == "-":
if sys.argv[i] in ["-h", "--help"]:
PrintHelp()
return 0
elif sys.argv[i] in ["-i", "--i", "-printid", "--printid"]:
isPrintID = True
i += 1
elif sys.argv[i] in ["-just-print-sum", "--just-print-sum"]:
isJustPrintSum = True
i += 1
elif sys.argv[i] in [ "-o", "--o", "-outfile", "--outfile"]:
outfile, i = myfunc.my_getopt_str(sys.argv, i)
elif sys.argv[i] in [ "-bs", "--bs", "-block-size", "--block-size"]:
BLOCK_SIZE, i = myfunc.my_getopt_int(sys.argv, i)
if BLOCK_SIZE < 0:
print >> sys.stderr,"Error! BLOCK_SIZE should >0"
return 1
else:
print >> sys.stderr,("Error! Wrong argument:%s" % sys.argv[i])
return 1
else:
infile=sys.argv[i]
i += 1
if myfunc.checkfile(infile) != 0:
return 1
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
status = Getseqlen(infile, isPrintID, isJustPrintSum, BLOCK_SIZE, fpout)
myfunc.myclose(fpout)
return status
def main(g_params):
# Check argv
numArgv = len(sys.argv)
if numArgv < 2:
PrintHelp()
return 1
outFile = ""
inFile = ""
begin = 0
end = 999999999
method = 2
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
isNonOptionArg = False
i = i + 1
elif sys.argv[i] == "--":
isNonOptionArg = True
i = i + 1
elif sys.argv[i][0] == "-":
if sys.argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif sys.argv[i] in ["-i", "--i", "--infile"]:
inFile, i = myfunc.my_getopt_str(sys.argv, i)
elif sys.argv[i] in ["-b", "--b", "--begin"]:
begin, i = myfunc.my_getopt_int(sys.argv, i)
elif sys.argv[i] in ["-e", "--e", "--end"]:
end, i = myfunc.my_getopt_int(sys.argv, i)
elif sys.argv[i] in ["-o", "--o", "--outfile"]:
outFile, i = my_getopt_str(sys.argv, i)
else:
print >> sys.stderr, ("Error! Wrong argument:%s" % sys.argv[i])
return 1
else:
inFile = sys.argv[i]
i += 1
if myfunc.checkfile(inFile, "Input file") != 0:
return 1
fpout = myfunc.myopen(outFile, sys.stdout, "w", False)
if method == 1:
CatFasta(inFile, begin, end, fpout)
else:
CatFasta2(inFile, begin, end, fpout)
myfunc.myclose(fpout)
return 0
def main(g_params):
# Check argv
numArgv=len(sys.argv)
if numArgv < 2:
PrintHelp()
return 1
outFile = ""
inFile = ""
begin = 0
end = 999999999
method = 2
i = 1
isNonOptionArg=False
while i < numArgv:
if isNonOptionArg == True:
isNonOptionArg=False
i = i + 1
elif sys.argv[i] == "--":
isNonOptionArg=True
i = i + 1
elif sys.argv[i][0] == "-":
if sys.argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif sys.argv[i] in ["-i", "--i", "--infile"]:
inFile, i = myfunc.my_getopt_str(sys.argv, i)
elif sys.argv[i] in ["-b", "--b", "--begin"]:
begin, i = myfunc.my_getopt_int(sys.argv, i)
elif sys.argv[i] in ["-e", "--e", "--end"]:
end, i = myfunc.my_getopt_int(sys.argv, i)
elif sys.argv[i] in ["-o" , "--o","--outfile"]:
outFile, i = my_getopt_str(sys.argv,i )
else:
print >> sys.stderr,("Error! Wrong argument:%s" % sys.argv[i])
return 1
else:
inFile=sys.argv[i]
i+=1
if myfunc.checkfile(inFile, "Input file") != 0:
return 1
fpout = myfunc.myopen(outFile, sys.stdout, "w", False)
if method == 1:
CatFasta(inFile,begin, end, fpout)
else:
CatFasta2(inFile,begin, end, fpout)
myfunc.myclose(fpout)
return 0
def main(): #{{{
numArgv = len(sys.argv)
if numArgv < 2:
PrintHelp()
return 1
outFile = ""
inFile = ""
N = 999999999
rand_seed = None
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
isNonOptionArg = False
i = i + 1
elif sys.argv[i] == "--":
isNonOptionArg = True
i = i + 1
elif sys.argv[i][0] == "-":
if sys.argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif sys.argv[i] in ["-i", "--infile"]:
inFile, i = myfunc.my_getopt_str(sys.argv, i)
elif sys.argv[i] in ["-n", "--n"]:
N, i = myfunc.my_getopt_int(sys.argv, i)
elif sys.argv[i] in ["-seed", "--seed"]:
rand_seed, i = myfunc.my_getopt_int(sys.argv, i)
elif sys.argv[i] in ["-o", "--outfile"]:
outFile, i = myfunc.my_getopt_str(sys.argv, i)
elif sys.argv[i] in ["-bs", "--block-size", "-block-size"]:
BLOCK_SIZE, i = myfunc.my_getopt_int(sys.argv, i)
if BLOCK_SIZE < 0:
print("Error! BLOCK_SIZE should >0", file=sys.stderr)
return 1
else:
print(("Error! Wrong argument:%s" % sys.argv[i]),
file=sys.stderr)
return 1
else:
inFile = sys.argv[i]
i += 1
if myfunc.checkfile(inFile, "Input file") != 0:
return 1
fpout = myfunc.myopen(outFile, sys.stdout, "w", False)
RandFasta(inFile, N, rand_seed, fpout)
myfunc.myclose(fpout)
def main():#{{{
numArgv=len(sys.argv)
if numArgv < 2:
PrintHelp()
return 1
outFile=""
inFile=""
N=999999999
rand_seed=None
i = 1
isNonOptionArg=False
while i < numArgv:
if isNonOptionArg == True:
isNonOptionArg=False
i = i + 1
elif sys.argv[i] == "--":
isNonOptionArg=True
i = i + 1
elif sys.argv[i][0] == "-":
if sys.argv[i] in [ "-h", "--help"]:
PrintHelp()
return 1
elif sys.argv[i] in ["-i", "--infile"]:
inFile, i = myfunc.my_getopt_str(sys.argv, i)
elif sys.argv[i] in [ "-n" , "--n"]:
N,i = myfunc.my_getopt_int(sys.argv, i)
elif sys.argv[i] in [ "-seed" , "--seed"]:
rand_seed, i = myfunc.my_getopt_int(sys.argv, i)
elif sys.argv[i] in [ "-o" , "--outfile"]:
outFile,i = myfunc.my_getopt_str(sys.argv, i)
elif sys.argv[i] in [ "-bs" , "--block-size" , "-block-size"]:
BLOCK_SIZE, i = myfunc.my_getopt_int(sys.argv, i)
if BLOCK_SIZE < 0:
print >> sys.stderr,"Error! BLOCK_SIZE should >0"
return 1
else:
print >> sys.stderr,("Error! Wrong argument:%s" % sys.argv[i])
return 1
else:
inFile = sys.argv[i]
i+=1
if myfunc.checkfile(inFile, "Input file") != 0:
return 1
fpout = myfunc.myopen(outFile, sys.stdout, "w", False)
RandFasta(inFile, N, rand_seed, fpout)
myfunc.myclose(fpout)
def PrepareDataForPairaln(seqid1, seqid2, outpath): #{{{
topoalnfile = "%s/%s.topoaln.fa" % (g_params['datapath'],
g_params['basename'])
if myfunc.checkfile(topoalnfile, "topoalnfile") != 0:
return 1
# 1 seqid1-seqid2.topoaln.fa
ext_topoaln = ".topoaln.fa"
pair_topoalnfile = "%s/%s_%s.topoaln.fa" % (outpath, seqid1, seqid2)
cmd = [
"%s/selectPairaln.py" % (rundir), "-pairaln", topoalnfile, "-outpath",
outpath, "-ext", ext_topoaln, "-split", seqid1, seqid2
]
try:
subprocess.check_output(cmd)
except subprocess.CalledProcessError, e:
print e
def main(): #{{{
numArgv = len(sys.argv)
if numArgv < 2:
PrintHelp()
return 1
argv = sys.argv
outfile = ""
infile = ""
output_format = "mfa"
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
infile = sys.argv[i]
isNonOptionArg = False
i = i + 1
elif sys.argv[i] == "--":
isNonOptionArg = True
i = i + 1
elif sys.argv[i][0] == "-":
if sys.argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif sys.argv[i] in ["-o", "--o"]:
outfile, i = myfunc.my_getopt_str(argv, i)
elif sys.argv[i] in ["-of", "--of"]:
output_format, i = myfunc.my_getopt_str(argv, i)
else:
print >> sys.stderr, ("Error! Wrong argument:%s" % argv[i])
return 1
else:
infile = argv[i]
i += 1
if myfunc.checkfile(infile, "MSA file") != 0:
return 1
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
# detect the format of mpa files, the old format
MPA2MSA(infile, output_format, fpout)
myfunc.myclose(fpout)
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outpath = ""
infile = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
infile = argv[i]
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-outpath", "--outpath"]:
(outpath, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
infile = argv[i]
i += 1
if myfunc.checkfile(infile) != 0:
return 1
if outpath == "":
print >> sys.stderr, "outpath not set"
elif not os.path.exists(outpath):
os.system("mkdir -p %s" % (outpath))
SplitPfamFasta(infile, outpath)
def main(g_params):#{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
infile = ""
i = 1
isNonOptionArg=False
while i < numArgv:
if isNonOptionArg == True:
infile = argv[i]
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-o", "--o", "-outfile"]:
(outfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
infile = argv[i]
i += 1
if myfunc.checkfile(infile) != 0:
return 1
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
goancDict =
myfunc.myclose(fpout)
def main(): #{{{
numArgv = len(sys.argv)
if numArgv < 2:
PrintHelp()
return 1
argv = sys.argv
outfile = ""
infile = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
infile = sys.argv[i]
isNonOptionArg = False
i = i + 1
elif sys.argv[i] == "--":
isNonOptionArg = True
i = i + 1
elif sys.argv[i][0] == "-":
if sys.argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif sys.argv[i] in ["-o", "--o"]:
outfile, i = myfunc.my_getopt_str(argv, i)
else:
print >> sys.stderr, ("Error! Wrong argument:%s" % argv[i])
return 1
else:
infile = argv[i]
i += 1
if myfunc.checkfile(infile) != 0:
return 1
return ReWriteFasta(infile, outfile)
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
msafile = ""
topofile = ""
topodb = ""
isIgnoreBadseq = True
method_match = 1
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
print("Error! Wrong argument:", argv[i], file=sys.stderr)
return 1
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-o", "--o", "-outfile"]:
(outfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-msa", "--msa"]:
(msafile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-topo", "--topo"]:
(topofile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-m", "--m"]:
(method_match, i) = myfunc.my_getopt_int(argv, i)
if method_match not in [0, 1]:
print("method_match %d not in [0,1]" % method_match,
file=sys.stderr)
return 1
elif argv[i] in ["-topodb", "--topodb"]:
(topodb, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in [
"-ig", "--ig", "-ignore-badseq", "--ignore-badseq"
]:
(tmpss, i) = myfunc.my_getopt_str(argv, i)
if tmpss[0].lower() == "y":
isIgnoreBadseq = True
else:
isIgnoreBadseq = False
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
else:
print("Error! Wrong argument:", argv[i], file=sys.stderr)
return 1
else:
print("Error! Wrong argument:", argv[i], file=sys.stderr)
return 1
if myfunc.checkfile(msafile) != 0:
return 1
if topodb != "":
if not os.path.exists(topodb + '0.db'):
print("topodb %s does not exist" % (topodb), file=sys.stderr)
return 1
else:
return MatchMSATopo_using_topodb(msafile, topodb, isIgnoreBadseq,
method_match, outfile)
elif topofile != "":
if not os.path.exists(topofile):
print("topofile %s does not exist" % (topofile), file=sys.stderr)
return 1
else:
return MatchMSATopo_using_topofile(msafile, topofile,
isIgnoreBadseq, method_match,
outfile)
else:
print("neither topofile nor topodb is set. exit", file=sys.stderr)
return 1
return 0
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
if outpath == "":
print >> sys.stderr, "Error! outpath not set. Exit"
print usage_short
return 1
elif not os.path.exists(outpath):
try:
subprocess.check_output(["mkdir", "-p", outpath])
except subprocess.CalledProcessError, e:
print e
return 1
if myfunc.checkfile(pfamid2seqidFile, "pfamid2seqidFile") != 0:
return 1
if myfunc.checkfile("%s0.db" % topodb, "topodb") != 0:
return 1
if myfunc.checkfile("%s0.db" % seqdb, "seqdb") != 0:
return 1
pfamid2seqidDict = myfunc.ReadFam2SeqidMap(pfamid2seqidFile)
hdl_topo = myfunc.MyDB(topodb)
if not hdl_topo.failure:
idSet_topo = set(hdl_topo.indexedIDList)
else:
idSet_topo = set([])
def main(g_params):#{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outpath = ""
modelfile = ""
seqfile = ""
tmpdir = ""
email = ""
jobid = ""
isKeepFiles = "no"
isRepack = "yes"
targetlength = None
i = 1
isNonOptionArg=False
while i < numArgv:
if isNonOptionArg == True:
modelfile = argv[i]
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-outpath", "--outpath"]:
(outpath, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-tmpdir", "--tmpdir"] :
(tmpdir, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-jobid", "--jobid"] :
(jobid, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-fasta", "--fasta"] :
(seqfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-k", "--k"] :
(isKeepFiles, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-r", "--r"] :
(isRepack, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-t", "--t"] :
(targetlength, i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-baseurl", "--baseurl"] :
(g_params['base_www_url'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-email", "--email"] :
(email, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
elif argv[i] in ["-force", "--force"]:
g_params['isForceRun'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
modelfile = argv[i]
i += 1
if jobid == "":
print >> sys.stderr, "%s: jobid not set. exit"%(sys.argv[0])
return 1
if myfunc.checkfile(modelfile, "modelfile") != 0:
return 1
if outpath == "":
print >> sys.stderr, "outpath not set. exit"
return 1
elif not os.path.exists(outpath):
try:
subprocess.check_output(["mkdir", "-p", outpath])
except subprocess.CalledProcessError, e:
print >> sys.stderr, e
return 1
def main(g_params):#{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
infile = ""
progList = []
progListFile = ""
outpath = ""
i = 1
isNonOptionArg=False
while i < numArgv:
if isNonOptionArg == True:
infile = argv[i]
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-prog", "--prog"]:
tmpstr, i = myfunc.my_getopt_str(argv, i)
progList.append(tmpstr)
elif argv[i] in ["-gzip", "--gzip"]:
tmpstr, i = myfunc.my_getopt_str(argv, i)
if tmpstr.upper()[0] == "-":
print >> sys.stderr, "Bad argument, -gzip should be"\
" followed by yes or no"
return 1
elif tmpstr.upper()[0] == "Y":
g_params['isGzip'] = True
else:
g_params['isGzip'] = False
elif argv[i] in ["-num", "--num"]:
g_params['num_per_split'], i = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-proglist", "--proglist"]:
progListFile, i = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-outpath", "--outpath"]:
outpath, i = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q"]:
g_params['isQuiet'] = True; i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
infile = argv[i]
i += 1
if myfunc.checkfile(infile,"infile") != 0:
return 1
inputList = ReadInputList(infile) # [(filename, numseq)]
inputList = sorted(inputList, key=lambda x:x[1], reverse=False)
rtname_infile = os.path.basename(os.path.splitext(infile)[0])
# get progList
if len(progList) == 0 and progListFile == "":
progList = default_progList
else:
if progListFile != "":
tmp_list = myfunc.ReadIDList(progListFile)
if len(tmp_list) == 0:
print >> sys.stderr, "progListFile %s does not exist or empty"%(
progListFile)
return 1
else:
progList += tmp_list
if len(progList) == 0:
print >> sys.stderr, "progList is empty. exit"
return 1
if outpath != "" and not os.path.exists(outpath):
try:
subprocess.check_output(["mkdir", "-p", outpath])
except subprocess.CalledProcessError, e:
print e
return 1
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outpath = ""
modelfile = ""
seqfile = ""
tmpdir = ""
email = ""
jobid = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
modelfile = argv[i]
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-outpath", "--outpath"]:
(outpath, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-tmpdir", "--tmpdir"]:
(tmpdir, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-jobid", "--jobid"]:
(jobid, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-fasta", "--fasta"]:
(seqfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-baseurl", "--baseurl"]:
(g_params['base_www_url'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-email", "--email"]:
(email, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
elif argv[i] in ["-force", "--force"]:
g_params['isForceRun'] = True
i += 1
else:
print("Error! Wrong argument:", argv[i], file=sys.stderr)
return 1
else:
modelfile = argv[i]
i += 1
if jobid == "":
print("%s: jobid not set. exit" % (sys.argv[0]), file=sys.stderr)
return 1
if myfunc.checkfile(modelfile, "modelfile") != 0:
return 1
if outpath == "":
print("outpath not set. exit", file=sys.stderr)
return 1
elif not os.path.exists(outpath):
try:
subprocess.check_output(["mkdir", "-p", outpath])
except subprocess.CalledProcessError as e:
print(e, file=sys.stderr)
return 1
if tmpdir == "":
print("tmpdir not set. exit", file=sys.stderr)
return 1
elif not os.path.exists(tmpdir):
try:
subprocess.check_output(["mkdir", "-p", tmpdir])
except subprocess.CalledProcessError as e:
print(e, file=sys.stderr)
return 1
g_params['debugfile'] = "%s/debug.log" % (outpath)
if not os.path.exists(path_profile_cache):
os.makedirs(path_profile_cache)
return RunJob(modelfile, seqfile, outpath, tmpdir, email, jobid, g_params)
def main(g_params):#{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outpath = ""
infile = ""
tmpdir = ""
email = ""
jobid = ""
i = 1
isNonOptionArg=False
while i < numArgv:
if isNonOptionArg == True:
infile = argv[i]
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-outpath", "--outpath"]:
(outpath, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-tmpdir", "--tmpdir"] :
(tmpdir, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-jobid", "--jobid"] :
(jobid, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-baseurl", "--baseurl"] :
(g_params['base_www_url'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-email", "--email"] :
(email, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
infile = argv[i]
i += 1
if jobid == "":
print >> sys.stderr, "%s: jobid not set. exit"%(sys.argv[0])
return 1
if myfunc.checkfile(infile, "infile") != 0:
return 1
if outpath == "":
print >> sys.stderr, "outpath not set. exit"
return 1
elif not os.path.exists(outpath):
try:
subprocess.check_output(["mkdir", "-p", outpath])
except subprocess.CalledProcessError, e:
print >> sys.stderr, e
return 1
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outpath = "./"
outfile = ""
idListFile = ""
uniprotDBname = ""
idList = []
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
idList.append(argv[i])
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-o", "--o", "-outfile"]:
(outfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-outpath", "--outpath"]:
(outpath, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-l", "--l"]:
(idListFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-uniprotdb", "--uniprotdb"]:
(uniprotDBname, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
idList.append(argv[i])
i += 1
if idListFile != "":
idList += myfunc.ReadIDList(idListFile)
if uniprotDBname == "":
print >> sys.stderr, "uniprotdb not set"
return 1
uniprotdbfile = "%s0.db" % uniprotDBname
if myfunc.checkfile(uniprotdbfile, "uniprotdbfile") != 0:
return 1
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
hdl = myfunc.MyDB(uniprotDBname)
if hdl.failure:
return 1
for seqid in idList:
data = hdl.GetRecord(seqid)
if data != None:
goinfo = GetGOInfoFromUniprotData(data)
WriteGOInfo(seqid, goinfo, fpout)
hdl.close()
myfunc.myclose(fpout)
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
fileListFile = ""
fileList = []
pfamDefFile = "%s/data/pfam/pfam26.0/Pfam-A.clans.tsv" % (DATADIR3)
threshold_Fraction_Group_2 = 0.05
threshold_NumSeq_Group_2 = 2
tableinfoFile = ""
pdbtospFile = ""
sprotACListFile = ""
threshold_g12_seqidt = 20.0
topoalnFile = ""
aapath = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
fileList.append(argv[i])
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-o", "--o", "-outfile"]:
(outfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-l", "--l"]:
(fileListFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-seqidttype", "--seqidttype"]:
(g_params['seqidttype'], i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-tableinfo", "--tableinfo"]:
(tableinfoFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-topoaln", "--topoaln"]:
(topoalnFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-aapath", "--aapath"]:
(aapath, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-thncls2", "--thncls2"]:
(threshold_NumSeq_Group_2, i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-thfrac2", "--thfrac2"]:
(threshold_Fraction_Group_2,
i) = myfunc.my_getopt_float(argv, i)
elif argv[i] in ["-pfamdef", "--pfamdef"]:
(pfamDefFile, i) = myfunc.my_getopt_str(argv, i)
elif (argv[i] in ["-pdbtosp", "--pdbtosp"]):
pdbtospFile, i = myfunc.my_getopt_str(argv, i)
elif (argv[i] in ["-sprot", "--sprot"]):
sprotACListFile, i = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
fileList.append(argv[i])
i += 1
if fileListFile != "":
fileList += myfunc.ReadIDList(fileListFile)
if len(fileList) < 1:
print >> sys.stderr, "No input set. exit"
return 1
if myfunc.checkfile(topoalnFile, "topoalnFile") != 0:
return 1
if myfunc.checkfile(aapath, "aapath") != 0:
return 1
if outfile == "":
print >> sys.stderr, "outfile not set. Exit"
return 1
outpath = myfunc.my_dirname(outfile)
if not os.path.exists(outpath):
cmd = ["mkdir", "-p", outpath]
try:
subprocess.check_output(cmd)
except subprocess.CalledProcessError, e:
print e
return 1
outfile_selected_pair, "-split", "-outpath", outpath, "-ext",
ext_topoaln
]
print '\n', " ".join(cmd), '\n'
try:
subprocess.check_output(cmd)
except subprocess.CalledProcessError, e:
print e
return 1
# draw pairwise topology alignment
for li in selectedPairList:
single_topoaln_file = "%s%s%s%s" % (outpath, os.sep, "%s_%s" %
(li[0], li[1]), ext_topoaln)
aafile = "%s%s%s%s" % (aapath, os.sep, li[3], ".fa")
if myfunc.checkfile(single_topoaln_file, "single_topoaln_file") != 0:
continue
if myfunc.checkfile(aafile, "aafile") != 0:
continue
method_shrink = "2"
method_plot = "mat"
shrinkrateTM = "3"
maxHoldLoop = "4"
cmd = [
"%s/drawMSATopo.py" % (binpath), single_topoaln_file, "-aaseq",
aafile, "-pdg", "yes", "-shrink", "yes", "-m-shrink",
method_shrink, "-method", method_plot, "-text", "yes", "-krbias",
"-shrinkrateTM", shrinkrateTM, "-max-hold-loop", maxHoldLoop
]
try:
subprocess.check_output(cmd)
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
pairalnFile = ""
pair = []
listfile = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
pair.append(argv[i])
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-o", "--o"]:
outfile, i = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-pairaln", "--pairaln"]:
pairalnFile, i = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-outpath", "--outpath"]:
g_params['outpath'], i = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-ext", "--ext"]:
g_params['ext'], i = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-l", "--l"]:
listfile, i = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q"]:
g_params['isQuiet'] = True
i += 1
elif argv[i] in ["-split", "--split"]:
g_params['isSplit'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
pair.append(argv[i])
i += 1
if myfunc.checkfile(pairalnFile, "pairalnFile") != 0:
return 1
if g_params['isSplit']:
if g_params['outpath'] == "":
print >> sys.stderr, "Error! outpath string is empty when 'split'"\
" is enabled. exit"
return 1
elif not os.path.exists(g_params['outpath']):
cmd = ["mkdir", "-p", g_params['outpath']]
subprocess.check_call(cmd)
pairlist = []
if len(pair) >= 2:
pairlist.append((pair[0], pair[1]))
if listfile != "":
pairlist += myfunc.ReadPairList(listfile)
pairlistSet = set([])
for pair in pairlist:
pairlistSet.add(pair)
del pairlist
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
SelectPairaln(pairalnFile, pairlistSet, fpout)
myfunc.myclose(fpout)
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outpath = ""
infile = ""
classfile = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
infile = argv[i]
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-outpath", "--outpath"]:
(outpath, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-class", "--class"]:
(classfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
infile = argv[i]
i += 1
if myfunc.checkfile(infile) != 0:
return 1
if myfunc.checkfile(classfile, "Class File") != 0:
return 1
if outpath == "":
outpath = os.path.dirname(infile)
if outpath == "":
outpath = "."
(id2ClassDict, classList) = ReadClassDict(classfile)
idList = myfunc.ReadIDList(infile)
rootname = os.path.basename(os.path.splitext(infile)[0])
ext = os.path.splitext(infile)[1]
fpoutList = {}
for i in range(len(classList)):
outfile = outpath + os.sep + rootname + ".%s" % classList[i] + ext
fpoutList[classList[i]] = open(outfile, "w")
for idd in idList:
try:
cls = id2ClassDict[idd]
except:
print >> sys.stderr, "id %s not in classDict" % idd
continue
fpoutList[cls].write("%s\n" % idd)
for i in range(len(classList)):
fpoutList[classList[i]].close()
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outpath = "./"
outfile = ""
real_topofile = ""
seqfile = ""
restrictIDListFile = ""
outfile_wrong_predtopo = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-o", "--o", "-outfile"]:
(outfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-owrong", "--owrong"]:
(outfile_wrong_predtopo, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-realtopo", "--realtopo"]:
(real_topofile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-seqfile", "--seqfile"]:
(seqfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-mode", "--mode"]:
(g_params['mode'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-path_predtopo", "--path_predtopo"]:
(g_params['path_predtopo'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-basename", "--basename"]:
(g_params['basename'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-restrictidlist", "--restrictidlist"]:
(restrictIDListFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
elif argv[i] in ["-rmsp", "--rmsp"]:
g_params['isRMSP'] = True
i += 1
elif argv[i] in ["-debug", "--debug"]:
g_params['isDEBUG'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
i += 1
if myfunc.checkfile(g_params['path_predtopo'], "path_predtopo") != 0:
return 1
if g_params['basename'] == "":
print >> sys.stderr, "%s: basename not set. exit" % (argv[0])
return 1
if myfunc.checkfile(real_topofile, "real_topofile") != 0:
return 1
if restrictIDListFile != "":
g_params['restrictIDset'] = set(myfunc.ReadIDList(restrictIDListFile))
g_params['isRestrictIDList'] = True
if g_params['mode'] == "":
if g_params['path_predtopo'].find("topcons_single") >= 0:
g_params['mode'] = "tps"
elif g_params['path_predtopo'].find("topcons") >= 0:
g_params['mode'] = "tp"
else:
print >> sys.stderr, "mode not set, and can not be recognized from path_predtopo=%s" % (
path_predtopo)
return 1
if not g_params['mode'] in ["tp", "tps"]:
print >> sys.stderr, "Unrecognized mode = %s" % (g_params['mode'])
return 1
(real_idlist, real_annolist,
real_topolist) = myfunc.ReadFasta(real_topofile)
seqDict = {}
if seqfile != "" and os.path.exists(seqfile):
(seq_idlist, seq_annolist, seqlist) = myfunc.ReadFasta(seqfile)
for i in xrange(len(seq_idlist)):
seqDict[seq_idlist[i]] = seqlist[i]
if len(real_idlist) <= 0:
print >> sys.stderr, "Failed to read real_topofile %s" % (
real_topofile)
return 1
real_topodict = {}
for i in xrange(len(real_idlist)):
real_topodict[real_idlist[i]] = real_topolist[i]
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
fpout_wrong = myfunc.myopen(outfile_wrong_predtopo, None, "w", False)
idSet_single = set([])
idSet_multi = set([])
for seqid in real_topodict:
topo = real_topodict[seqid]
numTM = myfunc.CountTM(topo)
if numTM == 1:
idSet_single.add(seqid)
elif numTM > 1:
idSet_multi.add(seqid)
# print "len(real_topodict)", len(real_topodict)
# print "len(idSet_single)", len(idSet_single)
# print "len(idSet_multi)", len(idSet_multi)
#for TM_type in ["All_Alpha", "Single", "Multi"]:
for TM_type in ["All_Alpha"]:
if TM_type == "All_Alpha":
sub_real_topodict = real_topodict
else:
sub_real_topodict = {}
for seqid in real_topodict:
topo = real_topodict[seqid]
numTM = myfunc.CountTM(topo)
if TM_type == "Single" and numTM == 1:
sub_real_topodict[seqid] = topo
elif TM_type == "Multi" and numTM > 1:
sub_real_topodict[seqid] = topo
Benchmark(sub_real_topodict, idSet_single, idSet_multi, TM_type, fpout,
fpout_wrong, seqDict)
myfunc.myclose(fpout)
def Itol_Tree(pfamid, datapath, outpath):
#Create the Itol class
itl = Itol.Itol()
#Set the tree file
treefile = "%s%s%s%s"%(datapath , os.sep , pfamid ,
".TMpro.clustalo10.fasttree")
taxofile = "%s%s%s%s"%(datapath , os.sep , pfamid ,
"-taxonomy-file.txt")
TMdeffile = "%s%s%s%s"%(datapath , os.sep , pfamid ,
"tms.txt")
if not os.path.exists(treefile):
print >> sys.stderr, "tree file %s does not exist. Ignore" %(treefile)
return 1
if myfunc.checkfile(taxofile, "taxofile") != 0:
return 1
numLeave = len(open(taxofile, "r").readlines())
# t = Tree(treefile)
# leaves = t.get_leaves()
fontsize = GetFontSize(numLeave)
colordeffile = taxofile
branchlabelfile = ""
dataset1 = ""
dataset2 = ""
dataset3 = ""
dataset4 = ""
dataset1 = TMdeffile
#===================================
itl.add_variable('treeFile',treefile)
itl.add_variable('treeName', pfamid)
itl.add_variable('treeFormat','newick')
if os.path.exists(colordeffile):
itl.add_variable('colorDefinitionFile', colordeffile)
if os.path.exists(branchlabelfile):
itl.add_variable('branchLabelsFile', branchlabelfile)
if os.path.exists(dataset1):
itl.add_variable('dataset1File',dataset1)
itl.add_variable('dataset1Label','numTM_and_repeat')
itl.add_variable('dataset1Separator','comma')
itl.add_variable('dataset1Type','domains')
itl.add_variable('dataset1PreventOverlap','1')
itl.add_variable('dataset1Color','#FF0000')
# itl.add_variable('dataset1BarSizeMax','300')
#===================================
# itl.add_variable('dataset1File',dataset1)
# itl.add_variable('dataset1Label','numTM')
# itl.add_variable('dataset1Separator','comma')
# itl.add_variable('dataset1Type','simplebar')
# itl.add_variable('dataset1Color','#FF0000')
#===================================
if os.path.exists(dataset2):
itl.add_variable('dataset2File', dataset2)
itl.add_variable('dataset2Label', 'taxonomy')
itl.add_variable('dataset2Separator','comma')
itl.add_variable('dataset2Type','colorstrip')
itl.add_variable('dataset2StripWidth','300')
itl.add_variable('dataset2PreventOverlap','1')
itl.add_variable('dataset2ColoringType','both')
itl.add_variable('dataset2CirclesSpacing','100')
#===================================
if os.path.exists(dataset3):
itl.add_variable('dataset3File', dataset3)
itl.add_variable('dataset3Label', 'pfam')
itl.add_variable('dataset3Separator','tab')
itl.add_variable('dataset3Type','colorstrip')
# itl.add_variable('dataset3Type','ColorDefinitionFile')
# itl.add_variable('dataset3StripWidth','300')
# itl.add_variable('dataset3PreventOverlap','1')
# itl.add_variable('dataset3ColoringType','both')
# itl.add_variable('dataset3CirclesSpacing','100')
#===================================
if os.path.exists(dataset4):
itl.add_variable('dataset4File', dataset4)
itl.add_variable('dataset4Label', 'cluster')
itl.add_variable('dataset4Separator','comma')
itl.add_variable('dataset4Type','colorstrip')
itl.add_variable('dataset4StripWidth','200')
itl.add_variable('dataset4PreventOverlap','1')
itl.add_variable('dataset4ColoringType','both')
#itl.add_variable('dataset1BarSizeMax','1')
#===================================
# Check parameters
# itl.print_variables()
#Submit the tree
print ''
print 'Uploading the tree. This may take some time depending on how large the tree is and how much load there is on the itol server'
good_upload = itl.upload()
if good_upload == False:
print 'There was an error:'+itl.comm.upload_output
sys.exit(1)
#Read the tree ID
print 'Tree ID: '+str(itl.comm.tree_id)
#Read the iTOL API return statement
print 'iTOL output: '+str(itl.comm.upload_output)
#Website to be redirected to iTOL tree
print 'Tree Web Page URL: '+itl.get_webpage()
# Warnings associated with the upload
print 'Warnings: '+str(itl.comm.warnings)
#Export to pdf
print 'Exporting to pdf'
itol_exporter = itl.get_itol_export()
#itol_exporter = itolexport.ItolExport()
#itol_exporter.set_export_param_value('tree','18793532031912684633930')
itol_exporter.set_export_param_value('format', 'eps')
itol_exporter.set_export_param_value('displayMode',"circular")
itol_exporter.set_export_param_value('showBS',"0")
itol_exporter.set_export_param_value('fontSize',fontsize)
itol_exporter.set_export_param_value('alignLabels',"1")
itol_exporter.set_export_param_value('datasetList','dataset1,dataset2,dataset3,dataset4')
epsfile = outpath + os.sep + pfamid + '-itol.eps'
pdffile = outpath + os.sep + pfamid + '-itol.pdf'
jpgfile = outpath + os.sep + pfamid + '-itol.jpg'
svgfile = outpath + os.sep + pfamid + '-itol.svg'
itol_exporter.export(epsfile)
itol_exporter.export(svgfile)
os.system("epstopdf %s" % epsfile )
os.system("convert %s %s" % (epsfile, jpgfile) )
print 'exported tree to ', pdffile
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outpath = "./"
infileList = []
gomapfile = "/data3/wk/MPTopo/pfamAna_refpro/pfammap_from_uniprot/Pfam-A-full.seqfrompfamfasta.percentTMpro_scampi.perTM75_nseq20.Family.nr100.filter.fragmented.uniq.pfam.goinfowithancestor.txt"
gotermfile = "/data3/wk/MPTopo/pfamAna_refpro/GO_analysis/GO_term.txt"
anclevel = 2
gotype = "function"
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
infileList.append(argv[i])
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-outpath", "--outpath"]:
(outpath, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-gomap", "--gomap"]:
(gomapfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-goterm", "--goterm"]:
(gotermfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-mp", "--mp"]:
(g_params['pairwise_comparison_method'],
i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
infileList.append(argv[i])
i += 1
# print len(gomapfile), gomapfile
# lines = open(gomapfile, "r").readlines()
# print lines
if myfunc.checkfile(gomapfile, "GO map file") != 0:
return 1
if myfunc.checkfile(gotermfile, "GO Term file") != 0:
return 1
cmpclassList = []
if g_params['pairwise_comparison_method'] == 1:
cmpclassList = cmpclassList_method1
elif g_params['pairwise_comparison_method'] == 3:
cmpclassList = cmpclassList_method3
else:
print >> sys.stderr, "mp not in [1,3]. Exit"
return 1
numCmpClass = len(cmpclassList)
numInfile = len(infileList)
if numInfile < len(cmpclassList):
print >> sys.stderr, "input file less than len(cmpclassList)=%d" % (
len(cmpclassList))
goMapDict = ReadGOMap(gomapfile)
goTermDict = ReadGOTerm(gotermfile)
pairinfoDict = {}
for infile in infileList:
tag = ""
for cls in cmpclassList:
if infile.find(".%s." % (cls)) != -1:
tag = cls
break
if tag == "":
print >> sys.stderr, "bad infile %s" % (infile)
return 1
pairinfoDict[tag] = ReadPairInfo(infile)
for tag in pairinfoDict:
pairinfo = pairinfoDict[tag]
for j in xrange(len(pairinfo)):
tup = pairinfo[j]
ancList1 = GetAncenstorGOList_LevelOne(tup[0], goMapDict, gotype)
ancList2 = GetAncenstorGOList_LevelOne(tup[1], goMapDict, gotype)
common_ancList = list(set(ancList1) & set(ancList2))
for goid in common_ancList:
pairinfo[j].append(goid)
# output pairinfo with GO common term
stemname = os.path.splitext(infileList[0].replace(".pairinfo.txt", ""))[0]
for tag in pairinfoDict:
outfile = outpath + os.sep + stemname + ".%s" % (
tag) + ".pairinfowithGOterm.txt"
WritePairInfoWithGO(pairinfoDict[tag], outfile)
print "%s output" % (outfile)
tableCmpclassDict = {}
tableNumTMHeatMapDict = {}
for goid in SEL_GOID_SET:
tableCmpclassDict[goid] = {}
tableNumTMHeatMapDict[goid] = {}
InitTableCmpClass(tableCmpclassDict[goid], numSeqIDTGroup, numCmpClass)
InitTableNumTMHeatMap(tableNumTMHeatMapDict[goid],
classList_TableNumTMHeatMap, MAX_NUMTM)
for tag in pairinfoDict.keys():
cmpclass = tag
idxClass = GetClassIndex(cmpclass, cmpclassList)
pairinfo = pairinfoDict[tag]
for li in pairinfo:
if len(li) > 9:
#print li
seqidt = li[8]
numTM1 = li[4]
numTM2 = li[5]
minNumTM = min(numTM1, numTM2)
maxNumTM = max(numTM1, numTM2)
for goid in li[9:]:
idxGroup = GetSeqIDTGroupIndex(seqidt, seqIDTGroup)
tableCmpclassDict[goid]['freq'][idxGroup][idxClass] += 1
tableCmpclassDict[goid]['subsum'][idxGroup] += 1
dt = tableNumTMHeatMapDict[goid]['ALL']
dt['data'][minNumTM][maxNumTM] += 1
if maxNumTM > dt['maxNumTM']:
dt['maxNumTM'] += 1
dt['numPair'] += 1
# write cmpclass
stemname2 = os.path.splitext(os.path.basename(stemname))[0]
print "stemname2=", stemname2
for goid in SEL_GOID_SET:
data = tableCmpclassDict[goid]
outfile = outpath + os.sep + stemname2 + "." + goid + ".cmpclass.txt"
if WriteTable2D(data['freq'], data['subsum'], cmpclassList,
seqIDTGroup, outfile) == 0:
xlabel = "Sequence identity"
print "%s output" % (outfile)
if g_params['pairwise_comparison_method'] == 1:
cmd = "%s/plotCmpClass_mp1_cmpsp_5.sh %s -xlabel \"%s\""\
" -outstyle eps -outpath %s -plot1 -multiplot" %(
binpath, outfile, xlabel, outpath)
elif g_params['pairwise_comparison_method'] == 3:
cmd = "%s/plotCmpClass_mp3.sh %s -xlabel \"%s\""\
" -outstyle eps -outpath %s -plot1 -multiplot" %(
binpath, outfile, xlabel, outpath)
os.system(cmd)
data = tableNumTMHeatMapDict[goid]
outfile = outpath + os.sep + stemname2 + "." + goid + ".numTMHeatMap.ALL.txt"
mtx = data['ALL']
mode_norm = "norm_diag"
print "%s numPair=%d" % (goid, mtx['numPair'])
if WriteNumTMHeatMap(mtx['data'], mtx['maxNumTM'], mtx['numPair'],
mode_norm, outfile) == 0:
cmd = "%s/plotNumTMHeatMap.sh %s" % (binpath, outfile)
os.system(cmd)
def main(g_params):#{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outpath = ""
infile = ""
tmpdir = ""
email = ""
jobid = ""
i = 1
isNonOptionArg=False
while i < numArgv:
if isNonOptionArg == True:
infile = argv[i]
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-outpath", "--outpath"]:
(outpath, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-tmpdir", "--tmpdir"] :
(tmpdir, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-jobid", "--jobid"] :
(jobid, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-baseurl", "--baseurl"] :
(g_params['base_www_url'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-email", "--email"] :
(email, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
infile = argv[i]
i += 1
if jobid == "":
print >> sys.stderr, "%s: jobid not set. exit"%(sys.argv[0])
return 1
if myfunc.checkfile(infile, "infile"):
return 1
if outpath == "":
print >> sys.stderr, "outpath not set. exit"
return 1
elif not os.path.exists(outpath):
try:
myfunc.check_output(["mkdir", "-p", outpath])
except subprocess.CalledProcessError, e:
print >> sys.stderr, e
return 1
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
infile = ""
gramPositiveFile = ""
gramNegativeFile = ""
eukFile = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
infile = argv[i]
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-o", "--o", "-outfile"]:
(outfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-gram+", "--gram+"]:
(gramPositiveFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-gram-", "--gram-"]:
(gramNegativeFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-euk", "--euk"]:
(eukFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
elif argv[i] in ["-debug", "--debug"]:
g_params['isDEBUG'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
infile = argv[i]
i += 1
if myfunc.checkfile(infile, "taxidwithtaxoFile") != 0:
return 1
if myfunc.checkfile(gramPositiveFile, "gramPositiveFile") != 0:
return 1
if myfunc.checkfile(gramNegativeFile, "gramNegativeFile") != 0:
return 1
if myfunc.checkfile(eukFile, "eukFile") != 0:
return 1
gramPositiveDict = ReadSignalPFile(gramPositiveFile)
gramNegativeDict = ReadSignalPFile(gramNegativeFile)
eukDict = ReadSignalPFile(eukFile)
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
hdl = myfunc.ReadLineByBlock(infile)
if hdl.failure:
return 1
lines = hdl.readlines()
while lines != None:
for line in lines:
strs = line.split("\t")
if len(strs) == 3:
seqid = strs[0].strip()
taxo = strs[2].strip()
info = ""
try:
if taxo == "Gram+" or taxo == "gram+":
info = gramPositiveDict[seqid]
elif taxo == "Gram-" or taxo == "gram-":
info = gramNegativeDict[seqid]
elif taxo == "Euk" or taxo == "euk":
info = eukDict[seqid]
if g_params['isDEBUG']:
print >> sys.stderr, "%s: %s" % (seqid, taxo)
except KeyError:
info = ""
if info != "":
fpout.write("%s\n" % info)
lines = hdl.readlines()
hdl.close()
myfunc.myclose(fpout)
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outpath = "./"
outfile = ""
infile = ""
goancfile = "%s%s%s" % (
DATADIR3, os.sep,
"wk/MPTopo/pfamAna_refpro/GO_analysis/GO_ancenstor.MF.txt")
uniqGOIDListFile = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
infile = argv[i]
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-o", "--o", "-outfile"]:
(outfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-wgoidlist", "--wgoidlist"]:
(uniqGOIDListFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-outpath", "--outpath"]:
(outpath, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-anc", "--anc"]:
(goancfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-l", "--l"]:
(idListFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-uniprotdb", "--uniprotdb"]:
(uniprotDBname, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
infile = argv[i]
i += 1
if myfunc.checkfile(infile) != 0:
return 1
if myfunc.checkfile(goancfile, "ancestor file") != 0:
return 1
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
GOinfoList = ReadGOInfo(infile)
GOAncDict = ReadGOAnc(goancfile)
if uniqGOIDListFile != "": #{{{
fp = open(uniqGOIDListFile, "w")
for goinfo in GOinfoList:
for item in ["function", "process", "location"]:
#for item in ["function"]:
if len(goinfo[item]) > 0:
for li in goinfo[item]:
goidSet.add(li[0])
for goid in goidSet:
print >> fp, goid
fp.close()
#}}}
for goinfo in GOinfoList:
nF = len(goinfo["function"])
nP = len(goinfo["process"])
nC = len(goinfo["location"])
fpout.write("%s F %1d P %1d C %1d\n" %
(goinfo['seqid'], nF, nP, nC))
for item in ["function", "process", "location"]:
for li in goinfo[item]:
try:
goancinfo = GOAncDict[li[0]]
except KeyError:
goancinfo = ""
fpout.write("%s\t%s\n" % (li[2], goancinfo))
myfunc.myclose(fpout)
def main(g_params):#{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outpath = ""
infile = ""
tmpdir = ""
email = ""
jobid = ""
i = 1
isNonOptionArg=False
while i < numArgv:
if isNonOptionArg == True:
infile = argv[i]
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-outpath", "--outpath"]:
(outpath, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-tmpdir", "--tmpdir"] :
(tmpdir, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-jobid", "--jobid"] :
(jobid, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-baseurl", "--baseurl"] :
(g_params['base_www_url'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-email", "--email"] :
(email, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
elif argv[i] in ["-force", "--force"]:
g_params['isForceRun'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
infile = argv[i]
i += 1
if jobid == "":
print >> sys.stderr, "%s: jobid not set. exit"%(sys.argv[0])
return 1
g_params['jobid'] = jobid
# create a lock file in the resultpath when run_job.py is running for this
# job, so that daemon will not run on this folder
lockname = "runjob.lock"
lock_file = "%s/%s/%s"%(path_result, jobid, lockname)
g_params['lockfile'] = lock_file
fp = open(lock_file, 'w')
try:
fcntl.lockf(fp, fcntl.LOCK_EX | fcntl.LOCK_NB)
except IOError:
print >> sys.stderr, "Another instance of %s is running"%(progname)
return 1
if myfunc.checkfile(infile, "infile") != 0:
return 1
if outpath == "":
print >> sys.stderr, "outpath not set. exit"
return 1
elif not os.path.exists(outpath):
cmd = ["mkdir", "-p", outpath]
(t_isCmdSuccess, t_runtime) = webcom.RunCmd(cmd, gen_logfile, gen_errfile)
if not t_isCmdSuccess:
return 1
if tmpdir == "":
print >> sys.stderr, "tmpdir not set. exit"
return 1
elif not os.path.exists(tmpdir):
cmd = ["mkdir", "-p", tmpdir]
(t_isCmdSuccess, t_runtime) = webcom.RunCmd(cmd, gen_logfile, gen_errfile)
if not t_isCmdSuccess:
return 1
numseq = myfunc.CountFastaSeq(infile)
g_params['debugfile'] = "%s/debug.log"%(outpath)
return RunJob(infile, outpath, tmpdir, email, jobid, g_params)
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
idListFile = ""
euk = ""
gram_pos = ""
gram_neg = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
idListFile = argv[i]
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-o", "--o", "-outfile"]:
(outfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-euk", "--euk"]:
(euk, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-gram+", "--gram+"]:
(gram_pos, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-gram-", "--gram-"]:
(gram_neg, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
idListFile = argv[i]
i += 1
if myfunc.checkfile(idListFile, "idListFile") != 0:
return 1
if myfunc.checkfile(euk, "euk") != 0:
return 1
if myfunc.checkfile(gram_pos, "gram_pos") != 0:
return 1
if myfunc.checkfile(gram_neg, "gram_neg") != 0:
return 1
idList = myfunc.ReadIDList(idListFile)
set_euk_idlist = set(myfunc.ReadIDList(euk))
set_gram_pos_idlist = set(myfunc.ReadIDList(gram_pos))
set_gram_neg_idlist = set(myfunc.ReadIDList(gram_neg))
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
NCBI_TaxID = ""
for i in xrange(len(idList)):
seqid = idList[i]
cls = ""
if seqid in set_euk_idlist:
cls = "euk"
elif seqid in set_gram_pos_idlist:
cls = "gram+"
elif seqid in set_gram_neg_idlist:
cls = "gram-"
else:
cls = "NA"
print >> fpout, "%s\t%s\t%s" % (seqid, NCBI_TaxID, cls)
myfunc.myclose(fpout)
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
workdir = ""
fileListFile = ""
idListFile = ""
extList = []
maxfile_per_folder = 2000
method = 0
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-idlist", "--idlist"]:
(idListFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-filelist", "--filelist"]:
(fileListFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-workdir", "--workdir"]:
(workdir, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-max", "--max"]:
(maxfile_per_folder, i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-method", "--method"]:
(method, i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-ext", "--ext"]:
(tmpstr, i) = myfunc.my_getopt_str(argv, i)
extList.append(tmpstr)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
if myfunc.checkfile(workdir) != 0:
return 1
if idListFile == "" and fileListFile == "":
print >> sys.stderr, "At least one of idListFile and fileListFile need to be set"
return 1
if idListFile != "":
if os.path.exists(idListFile):
idList = myfunc.ReadIDList(idListFile)
if len(idList) <= 0:
print >> sys.stderr, "No ID in idListFile %s" % (idListFile)
elif len(extList) <= 0:
print >> sys.stderr, "No extension set when idList is used."
else:
SplitToFolder_idlist(idList, workdir, extList,
maxfile_per_folder)
else:
print >> sys.stderr, "idListFile %s does not exist" % (idListFile)
if fileListFile != "":
if os.path.exists(fileListFile):
fileList = open(fileListFile, "r").read().split("\n")
fileList = filter(None, fileList)
if len(fileList) <= 0:
print >> sys.stderr, "No file in fileListFile %s" % (
fileListFile)
else:
SplitToFolder_filelist(fileList, workdir, maxfile_per_folder)
else:
print >> sys.stderr, "fileListFile %s does not exist" % (
fileListFile)
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
outfile_with_famid = ""
outfile_with_pdb = ""
outfile_fam2seqmap = ""
idListFile = ""
mapfile = "%s%s%s" % (
DATADIR3, os.sep,
"wk/MPTopo/pfamAna_refpro/pfammap_from_uniprot/refpro20120604-celluar.selmaxlength-m1.nr100.filter.fragmented.clanid2seqid"
)
restrictIDListFile = ""
idList = []
maxseq_for_fam = 200
maxpair_for_fam = 300
method = 0
rand_seed = None
pdbtospFile = ""
isOnlyPDB = False
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
idList.append(argv[i])
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-o", "--o", "-outfile", "--outfile"]:
outfile, i = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-outwithfamid", "--outwithfamid"]:
outfile_with_famid, i = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-outfam2seqmap", "--outfam2seqmap"]:
outfile_fam2seqmap, i = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-outwithpdb", "--outwithpdb"]:
outfile_with_pdb, i = myfunc.my_getopt_str(argv, i)
elif argv[i] in [
"-tmprolist", "--tmprolist", "-restrictlist",
"--restrictlist"
]:
restrictIDListFile, i = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-mapfile", "--mapfile"]:
mapfile, i = myfunc.my_getopt_str(argv, i)
elif (argv[i] in ["-pdbtosp", "--pdbtosp"]):
pdbtospFile, i = myfunc.my_getopt_str(argv, i)
elif sys.argv[i] in ["-seed", "--seed"]:
rand_seed, i = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-l", "--l"]:
idListFile, i = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-maxseq", "--maxseq"]:
maxseq_for_fam, i = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-maxpair", "--maxpair"]:
maxpair_for_fam, i = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-m", "--m", "-method", "--method"]:
method, i = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-q"]:
g_params['isQuiet'] = True
i += 1
elif argv[i] in ["-onlypdb", "--onlypdb"]:
g_params['isOnlyPDB'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
idList.append(argv[i])
i += 1
if os.path.exists(idListFile):
idList += myfunc.ReadIDList(idListFile)
if len(idList) < 1:
print >> sys.stderr, "no ID set. exit"
return 1
if myfunc.checkfile(mapfile, "idMapFile") != 0:
return 1
idMapDict = myfunc.ReadFam2SeqidMap(mapfile)
# Read in pdbtosp map
if pdbtospFile != "":
(pdb2uniprotMap, uniprot2pdbMap) =\
myfunc.ReadPDBTOSP(pdbtospFile)
g_params['uniprotidlist_with_pdb'] = set(uniprot2pdbMap.keys())
g_params['uniprot2pdbMap'] = uniprot2pdbMap
if g_params['isOnlyPDB'] == True:
if pdbtospFile == "":
print >> sys.stderr, "onlypdb is true but pdbtospFile is not set. exit."
return 1
elif g_params['uniprotidlist_with_pdb'] == set([]):
print >> sys.stderr, "onlypdb is true but uniprotidlist_with_pdb is empty. exit."
return 1
restrictIDSet = set([])
if restrictIDListFile != "":
restrictIDSet = set(myfunc.ReadIDList(restrictIDListFile))
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
fpout_withfamid = myfunc.myopen(outfile_with_famid, None, "w", False)
fpout_withpdb = myfunc.myopen(outfile_with_pdb, None, "w", False)
fpout_fam2seqmap = myfunc.myopen(outfile_fam2seqmap, None, "w", False)
if method == 0:
GeneratePairWithinFam_m_0(idList, idMapDict, restrictIDSet,
maxseq_for_fam, rand_seed, fpout,
fpout_withfamid)
elif method == 1:
GeneratePairWithinFam_m_1(idList, idMapDict, restrictIDSet,
maxpair_for_fam, rand_seed, fpout,
fpout_withfamid, fpout_fam2seqmap)
elif method == 2: #all to all
GeneratePairWithinFam_m_2(idList, idMapDict, restrictIDSet, fpout,
fpout_withfamid, fpout_withpdb)
myfunc.myclose(fpout)
myfunc.myclose(fpout_withfamid)
myfunc.myclose(fpout_withpdb)
myfunc.myclose(fpout_fam2seqmap)
return 0
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
SPE_PAIR_LIST = [(2, 1), (2, 4), (2, 6), (2, 8), (3, 6), (3, 7), (4, 6),
(4, 8), (4, 10), (5, 7), (5, 10), (6, 8), (6, 10),
(6, 12), (7, 14), (8, 10), (8, 12), (10, 12), (10, 13),
(11, 13), (12, 14)]
outfile = ""
infile = ""
pfamDefFile = "%s/data/pfam/pfam26.0/Pfam-A.clans.tsv" % (DATADIR3)
signalpFile = "%s/wk/MPTopo/pfamAna_refpro/pred_signalp/refpro20120604-celluar.selmaxlength-m1.nr100.signalp_list" % (
DATADIR3)
#seqid2clanidMapFile = "%s/wk/MPTopo/pfamAna_refpro/pfammap_from_uniprot/refpro20120604-celluar.selmaxlength-m1.nr100.filter.fragmented.seqid2clanid"%(DATADIR3)
#seqid2pfamidMapFile = "%s/wk/MPTopo/pfamAna_refpro/pfammap_from_uniprot/refpro20120604-celluar.selmaxlength-m1.nr100.filter.fragmented.seqid2pfamid"%(DATADIR3)
seqid2clanidMapFile = ""
seqid2pfamidMapFile = ""
tm_pfamidListFile = ""
tm_clanidListFile = ""
pfamid2seqidMapFile = ""
clanid2seqidMapFile = ""
dbname_predTM = ""
pairlistwithpfamidFile = ""
pfamtype = ""
pairListFile = ""
#classList_TableNumTMHeatMap = ["ALL", "RMSP"]
classList_TableNumTMHeatMap = ["ALL"]
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
infile = argv[i]
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-o", "--o", "-outfile"]:
(outfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-outpath", "--outpath"]:
(g_params['outpath'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-l", "--l"]:
(fileListFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-pfamdef", "--pfamdef"]:
(pfamDefFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-signalp", "--signalp"]:
(signalpFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-mp", "--mp"]:
g_params[
'pairwise_comparison_method'], i = myfunc.my_getopt_int(
argv, i)
elif argv[i] in ["-mindiffpair", "--mindiffpair"]:
g_params['mindiffpair'], i = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-pfamtype", "--pfamtype"]:
pfamtype, i = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-clanidlist", "--clanidlist"]:
(tm_clanidListFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-pfamidlist", "--pfamidlist"]:
(tm_pfamidListFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-seqid2clanid", "--seqid2clanid"]:
(seqid2clanidMapFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-seqid2pfamid", "--seqid2pfamid"]:
(seqid2pfamidMapFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-pfamid2seqid", "--pfamid2seqid"]:
(pfamid2seqidMapFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-clanid2seqid", "--clanid2seqid"]:
(clanid2seqidMapFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-pairlistwithpfamid", "--pairlistwithpfamid"]:
(pairlistwithpfamidFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-predTMdbname", "--predTMdbname"]:
(dbname_predTM, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-pairlist", "--pairlist"]:
(pairListFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-winsize", "--winsize"]:
(g_params['winsize'], i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-outname", "--outname"]:
(g_params['outname'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
elif argv[i] in ["-prokar", "--prokar"]:
g_params['isOnlyAnaProkar'] = True
i += 1
elif argv[i] in ["-eukar", "--eukar"]:
g_params['isOnlyAnaEukar'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
infile = argv[i]
i += 1
if myfunc.checkfile(
infile, "%s (line %d): infile" %
(__file__, inspect.currentframe().f_lineno)) != 0:
return 1
dirpath = myfunc.my_dirname(infile)
# try to obtain Pfam family tag
tag = ""
if pfamtype != "":
if pfamtype.upper().find("FAM") != -1:
tag = ".Family"
elif pfamtype.upper().find("DOM") != -1:
tag = ".Domain"
elif pfamtype.upper().find("REP") != -1:
tag = ".Repeat"
elif pfamtype.upper().find("MOT") != -1:
tag = ".Motif"
else:
tag = ""
else:
if infile.find(".Family.") != -1:
tag = ".Family"
elif infile.find(".Domain.") != -1:
tag = ".Domain"
elif infile.find(".Repeat.") != -1:
tag = ".Repeat"
elif infile.find(".Motif.") != -1:
tag = ".Motif"
else:
tag = ""
if seqid2clanidMapFile == "":
seqid2clanidMapFile = "%s/wk/MPTopo/pfamAna_refpro/pfammap_from_uniprot/Pfam-A-full.seqfrompfamfasta.percentTMpro_scampi.perTM75_nseq20.nr100.filter.fragmented.seqid2clanid" % (
DATADIR3)
if myfunc.checkfile(
seqid2clanidMapFile, "%s (line %d): seqid2clanidMapFile" %
(__file__, inspect.currentframe().f_lineno)):
return 1
if seqid2pfamidMapFile == "":
seqid2pfamidMapFile = "%s/wk/MPTopo/pfamAna_refpro/pfammap_from_uniprot/Pfam-A-full.seqfrompfamfasta.percentTMpro_scampi.perTM75_nseq20%s.nr100.filter.fragmented.seqid2pfamid" % (
DATADIR3, tag)
if myfunc.checkfile(
seqid2pfamidMapFile, "%s (line %d): seqid2pfamidMapFile" %
(__file__, inspect.currentframe().f_lineno)):
return 1
if pfamid2seqidMapFile == "":
pfamid2seqidMapFile = "%s/wk/MPTopo/pfamAna_refpro/pfammap_from_uniprot/Pfam-A-full.seqfrompfamfasta.percentTMpro_scampi.perTM75_nseq20.nr100.filter.fragmented.pfamid2seqid" % (
DATADIR3)
if myfunc.checkfile(
pfamid2seqidMapFile, "%s (line %d): pfamid2seqidMapFile" %
(__file__, inspect.currentframe().f_lineno)):
return 1
if clanid2seqidMapFile == "":
clanid2seqidMapFile = "%s/wk/MPTopo/pfamAna_refpro/pfammap_from_uniprot/Pfam-A-full.seqfrompfamfasta.percentTMpro_scampi.perTM75_nseq20%s.nr100.filter.fragmented.clanid2seqid" % (
DATADIR3, tag)
if myfunc.checkfile(
clanid2seqidMapFile, "%s (line %d): clanid2seqidMapFile" %
(__file__, inspect.currentframe().f_lineno)):
return 1
if tm_pfamidListFile == "":
tm_pfamidListFile = "%s/data/pfam/pfam26.0/Pfam-A-full.seqfrompfamfasta.percentTMpro_scampi.perTM75_nseq20%s.pfamidlist" % (
DATADIR3, tag)
if myfunc.checkfile(
tm_pfamidListFile, "%s (line %d): tm_pfamidListFile" %
(__file__, inspect.currentframe().f_lineno)):
return 1
if tm_clanidListFile == "":
tm_clanidListFile = "%s/data/pfam/pfam26.0/Pfam-A-full.seqfrompfamfasta.percentTMpro_scampi.perTM75_nseq20.clanidlist" % (
DATADIR3)
if myfunc.checkfile(
tm_clanidListFile, "%s (line %d): tm_clanidListFile" %
(__file__, inspect.currentframe().f_lineno)):
return 1
if dbname_predTM == "":
dbname_predTM = "%s/wk/MPTopo/pfamAna_refpro/pred_topcons_single_method4/refpro20120604-celluar.selmaxlength-m1.topcons-single_topcons_single.m1.agree-44.RMSP" % (
DATADIR3)
if myfunc.checkfile(
"%s0.db" % (dbname_predTM), "%s (line %d): dbname_predTM" %
(__file__, inspect.currentframe().f_lineno)):
return 1
if g_params['isOnlyAnaProkar']:
prokarseqidfile = "%s/data/uniprot/reference_proteome/refpro20120604-celluar.selmaxlength-m1.nr100.filter.fragmented.Prokaryota.seqidlist" % (
DATADIR3)
g_params['prokarSeqIDSet'] = set(myfunc.ReadIDList(prokarseqidfile))
if len(g_params['prokarSeqIDSet']) < 1:
return 1
if g_params['isOnlyAnaEukar']:
eukarseqidfile = "%s/data/uniprot/reference_proteome/refpro20120604-celluar.selmaxlength-m1.nr100.filter.fragmented.Eukaryota.seqidlist" % (
DATADIR3)
g_params['eukarSeqIDSet'] = set(myfunc.ReadIDList(eukarseqidfile))
if len(g_params['eukarSeqIDSet']) < 1:
return 1
if pairlistwithpfamidFile == "":
pairlistwithpfamidFile = "%s/../../Pfam-.maxpair100.pairlistwithpfamid" % (
dirpath)
if myfunc.checkfile(
pairlistwithpfamidFile, "%s (line %d): pairlistwithpfamidFile" %
(__file__, inspect.currentframe().f_lineno)):
return 1
pfamid_2_seqidpair_Dict = ReadPairListWithFamID(pairlistwithpfamidFile)
usedPfamIDSet = set(
pfamid_2_seqidpair_Dict.keys()) # pfamids used in pair selection
if pairListFile != "":
li = myfunc.ReadPairList(pairListFile)
SPE_PAIR_LIST = []
for tup in li:
SPE_PAIR_LIST.append((int(tup[0]), int(tup[1])))
(pfamidDefDict, clanidDefDict) = ReadPfamDefFile(pfamDefFile)
signalpDict = lcmp.ReadSignalPDict(signalpFile)
seqid2clanidDict = myfunc.ReadFam2SeqidMap(seqid2clanidMapFile)
seqid2pfamidDict = myfunc.ReadFam2SeqidMap(seqid2pfamidMapFile)
clanid2seqidDict = myfunc.ReadFam2SeqidMap(clanid2seqidMapFile)
pfamid2seqidDict = myfunc.ReadFam2SeqidMap(pfamid2seqidMapFile)
tm_pfamidList = myfunc.ReadIDList(tm_pfamidListFile)
tm_clanidList = myfunc.ReadIDList(tm_clanidListFile)
tm_pfamidSet = set(tm_pfamidList)
tm_clanidSet = set(tm_clanidList)
hdl_predTM = myfunc.MyDB(dbname_predTM)
if not hdl_predTM.failure:
idSet_TMpro = set(hdl_predTM.indexedIDList)
else:
idSet_TMpro = set([])
#classList_TableNumTMHeatMap = ["ALL", "RMSP", "RMDUP"]
#alignrangeList = ['FULL_ALIGNED', 'all', 'PART_ALIGNED']
alignrangeList = ['FULL_ALIGNED']
if g_params['outpath'] != "" and not os.path.exists(g_params['outpath']):
cmd = ["mkdir", "-p", g_params['outpath']]
try:
subprocess.check_call(cmd)
except subprocess.CalledProcessError, e:
print e
return 1