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 main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
datapath = "."
outpath = './'
idList = []
idListFile = ''
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
isNonOptionArg = False
idList.append(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 1
elif sys.argv[i] in ["-datapath", "--datapath"]:
datapath = sys.argv[i + 1]
i += 2
elif argv[i] in ["-method", "--method"]:
g_params['method'], i = myfunc.my_getopt_int(sys.argv, i)
elif sys.argv[i] in ["-l", "--l"]:
idListFile = sys.argv[i + 1]
i = i + 2
elif sys.argv[i] in ["-outpath", "--outpath"]:
outpath = sys.argv[i + 1]
i = i + 2
else:
print >> sys.stderr, ("Error! Wrong argument:%s" % sys.argv[i])
return 1
else:
idList.append(sys.argv[i])
i += 1
if idListFile != "":
idList += myfunc.ReadIDList(idListFile)
if len(idList) > 0:
os.system("mkdir -p %s" % outpath)
cnt = 0
for pfamid in idList:
print "================== ", cnt, pfamid, " ===================="
if g_params['method'] == 0:
Itol_Tree_m0(pfamid, datapath, outpath)
elif g_params['method'] == 1:
Itol_Tree_m1(pfamid, datapath, outpath)
cnt += 1
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 = ""
fileList = []
fileListFile = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
fileList.append(sys.argv[i])
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 sys.argv[i] in ["-l", "--l"]:
fileListFile, i = myfunc.my_getopt_str(sys.argv, i)
elif sys.argv[i] in ["-o", "--o", "-outfile"]:
outfile, i = myfunc.my_getopt_str(sys.argv, i)
elif sys.argv[i] in ["-evalue", "--evalue"]:
g_params['evalue_th'], i = myfunc.my_getopt_float(sys.argv, i)
elif sys.argv[i] in ["-seqidt", "--seqidt"]:
g_params['seqidt_th'], i = myfunc.my_getopt_float(sys.argv, i)
elif sys.argv[i] in ["-round", "--round"]:
g_params['iteration'] = myfunc.my_getopt_int(sys.argv, i)
else:
print >> sys.stderr, ("Error! Wrong argument: '%s'" %
sys.argv[i])
return 1
else:
fileList.append(sys.argv[i])
i += 1
if fileListFile != "":
fileList += myfunc.ReadIDList(fileListFile, delim="\n")
if len(fileList) < 1:
print >> sys.stderr, "No input set. exit"
return 1
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
for infile in fileList:
BlastM9toPairlist(infile, 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 >> 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): #{{{
argv = sys.argv
outpath = ''
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
fileList = []
method = 0
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
isNonOptionArg = False
fileList.append(argv[i])
i = i + 1
elif argv[i] == "--":
isNonOptionArg = True
i = i + 1
elif argv[i][0] == "-":
if argv[i] == "-h" or argv[i] == "--help":
PrintHelp()
sys.exit(0)
elif argv[i] in ["-outpath", "--outpath"]:
(outpath, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-m", "--m"]:
(method, i) = myfunc.my_getopt_int(argv, i)
else:
print >> sys.stderr, ("Error! Wrong argument:%s" % argv[i])
return 1
else:
fileList.append(argv[i])
i += 1
if len(fileList) < 1:
print >> sys.stderr, "No input set"
return 1
cnt = 0
for treefile in fileList:
print "================== ", cnt, treefile, " ===================="
if method == 0:
Itol_Tree_m0(treefile, outpath, method)
elif method == 1:
Itol_Tree_m1(treefile, outpath, method)
cnt += 1
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outpath = ""
datapath = ""
pairListFile = ""
seqlenFile = ""
shortid2fullidFile = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
isNonOptionArg = False
i += 1
return 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 ["-datapath", "--datapath"]:
(datapath, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-treepath", "--treepath"]:
(g_params['treepath'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-maxperfamily", "--maxperfamily"]:
(g_params['max_num_output_per_family'],
i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-swissprot", "--swissprot"]:
(g_params['isSP_threshold'], i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-msapath", "--msapath"]:
(g_params['msapath'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-seqlen", "--seqlen"]:
(seqlenFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-shortid2fullid", "--shortid2fullid"]:
(shortid2fullidFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-l", "--l"]:
(pairListFile, 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:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
if pairListFile != "":
pairInfoDict = ReadPairInfo(pairListFile)
if len(pairInfoDict) < 1:
print >> sys.stderr, "no pair info read in. exit"
return 1
if datapath == "":
print >> sys.stderr, "datapath not set"
return 1
elif not os.path.exists(datapath):
print >> sys.stderr, "datapath %s does not exist" % (datapath)
return 1
if outpath == "":
print >> sys.stderr, "outpath not set"
return 1
elif not os.path.exists(outpath):
cmd = ["mkdir", "-p", outpath]
subprocess.check_call(cmd)
g_params['OS'] = os.uname()[0]
if g_params['OS'].find('Linux') != -1:
g_params['CP_EXE'] = "/bin/cp -uf"
else:
g_params['CP_EXE'] = "/bin/cp -f"
# read seqlen file
if seqlenFile != "":
g_params['seqLenDict'] = myfunc.ReadSeqLengthDict(seqlenFile)
if shortid2fullidFile != "":
g_params['uniprotAC2FullSeqIDMap'] = myfunc.ReadID2IDMap(
shortid2fullidFile)
#Sort by descending order of numseq
pairInfoList = sorted(pairInfoDict.items(),
key=lambda x: x[1]['numseq'],
reverse=True)
#pairInfoList ['pfamid', {'numseq': 123, 'pairlist':[]}]
WriteHTML(pairInfoList, datapath, outpath)
def main(g_params):#{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
rmsg = ""
outpath = ""
jobid = ""
datapath = ""
numseq = -1
numseq_this_user = -1
email = ""
host_ip = ""
base_www_url = ""
i = 1
isNonOptionArg=False
while i < numArgv:
if isNonOptionArg == True:
print >> g_params['fperr'], "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 ["-outpath", "--outpath"]:
(outpath, 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 ["-host", "--host"]:
(host_ip, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-nseq", "--nseq"]:
(numseq, i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-nseq-this-user", "--nseq-this-user"]:
(numseq_this_user, i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-baseurl", "--baseurl"]:
(base_www_url, 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 ["-datapath", "--datapath"] :
(datapath, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-force", "--force"]:
g_params['isForceRun'] = True
i += 1
elif argv[i] in ["-runlocal", "--runlocal"]:
g_params['isRunLocal'] = True
i += 1
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> g_params['fperr'], "Error! Wrong argument:", argv[i]
return 1
else:
print >> g_params['fperr'], "Error! Wrong argument:", argv[i]
return 1
if outpath == "":
print >> g_params['fperr'], "outpath not set. exit"
return 1
elif not os.path.exists(outpath):
cmd = ["mkdir", "-p", outpath]
try:
rmsg = subprocess.check_output(cmd, stderr=subprocess.STDOUT)
except subprocess.CalledProcessError, e:
print e
print rmsg
return 1
def main(g_params): #{{{
argv = sys.argv
numArgv = len(sys.argv)
if numArgv < 2:
PrintHelp()
return 1
infile = ""
outpath = "./"
isQuiet = False
tableinfoFile = ""
cmpclassList = []
restrictIDListFile = ""
signalpFile = ""
dupFile = ""
outfile = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
infile = sys.argv[i]
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()
sys.exit()
elif argv[i] in ["-o", "--o"]:
(outfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-cmpclass", "--cmpclass"]:
(tmpstr, i) = myfunc.my_getopt_str(argv, i)
cmpclassList.append(tmpstr)
elif argv[i] in ["-signalp", "--signalp"]:
(signalpFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-restrictidlist", "--restrictidlist"]:
(restrictIDListFile, i) = myfunc.my_getopt_str(argv, i)
g_params['isRestrictIDListSet'] = True
elif argv[i] in ["-dup", "--dup", "-dupfile", "--dupfile"]:
(dupFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-rmsp", "--rmsp"]:
g_params['isRemoveSignalP'] = True
i += 1
elif argv[i] in ["-rmdup", "--rmdup"]:
g_params['isRemoveDup'] = True
i += 1
elif argv[i] in ["-seq2fammap", "--seq2fammap"]:
(seq2famMapfile, 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 ["-min-seqidt", "--min-seqidt"]:
g_params['minSeqIDT'], i = myfunc.my_getopt_float(argv, i)
elif argv[i] in ["-max-seqidt", "--max-seqidt"]:
g_params['maxSeqIDT'], i = myfunc.my_getopt_float(argv, i)
elif argv[i] in ["-evodist", "--evodist"]:
g_params['isEvodist'] = True
i += 1
elif argv[i] in ["-alignrange", "--alignrange"]:
g_params['alignrange'], i = myfunc.my_getopt_str(argv, i)
if not g_params['alignrange'] in ['all', 'full', 'part']:
print >> sys.stderr, "alignrange must be one of [all, full, part]"
return 1
else:
if g_params['alignrange'] == 'full':
g_params['alignrange'] = 'FULL_ALIGNED'
elif g_params['alignrange'] == 'part':
g_params['alignrange'] = 'PART_ALIGNED'
elif argv[i] in ["-debug", "--debug"]:
if argv[i + 1][0].lower() == 'y':
g_params['isDEBUG'] = True
else:
g_params['isDEBUG'] = False
i += 2
elif argv[i] in [
"-debug-unmapped-position", "--debug-unmapped-position"
]:
DEBUG_UNMAPPED_TM_POSITION = 1
i += 2
elif sys.argv[i] == "-q":
isQuiet = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", sys.argv[i]
return -1
else:
infile = sys.argv[i]
i += 1
if infile == "":
print >> sys.stderr, "infile not set. Exit."
return -1
elif not os.path.exists(infile):
print >> sys.stderr, "infile %s does not exists. Exit." % infile
try:
fpin = open(infile, "rb")
except IOError:
print >> sys.stderr, "Failed to open input file %s" % (infile)
return -1
pairalnStat = {}
if g_params['seqidttype'] != 0:
if tableinfoFile == "" or not os.path.exists(tableinfoFile):
print >> sys.stderr, "tableinfoFile must be set when seqidttype is set to 1 or 2"
print >> sys.stderr, "but seqidttype = %d is set. Exit." % g_params[
'seqidttype']
return -1
pairalnStat = lcmp.ReadPairAlnTableInfo(tableinfoFile)
rootname = os.path.basename(os.path.splitext(infile)[0])
binpath = os.path.dirname(sys.argv[0])
signalpDict = {}
if signalpFile != "":
signalpDict = lcmp.ReadSignalPDict(signalpFile)
if signalpDict != {}:
g_params['isSignalPSet'] = True
dupPairList = []
if dupFile != "":
dupPairList = lcmp.ReadDupPairList(dupFile)
if len(dupPairList) > 0:
g_params['isDupSet'] = True
dupPairSet = set(dupPairList)
restrictIDSet = set([])
if restrictIDListFile != "":
restrictIDSet = set(myfunc.ReadIDList(restrictIDListFile))
rltyDict = {}
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
unprocessedBuffer = ""
cntTotalReadInRecord = 0
cntTotalOutputRecord = 0
isEOFreached = False
while 1:
buff = fpin.read(BLOCK_SIZE)
if buff == "":
isEOFreached = True
buff = unprocessedBuffer + buff
pairCmpRecordList = []
unprocessedBuffer = lcmp.ReadPairCmpResultFromBuffer(
buff, pairCmpRecordList)
AddTableInfo(pairCmpRecordList, pairalnStat)
AddSignalPInfo(pairCmpRecordList, signalpDict)
AddDupInfo(pairCmpRecordList, dupPairSet)
cntTotalReadInRecord += len(pairCmpRecordList)
pairCmpRecordList = FilterPairCmpResult(pairCmpRecordList,
cmpclassList, rltyDict,
restrictIDSet)
if len(pairCmpRecordList) > 0:
lcmp.WritePairCmpRecord(pairCmpRecordList, cntTotalOutputRecord,
fpout)
cntTotalOutputRecord += len(pairCmpRecordList)
if isEOFreached == True:
break
fpin.close()
print "cntTotalReadInRecord =", cntTotalReadInRecord
print "cntTotalOutputRecord =", cntTotalOutputRecord
myfunc.myclose(fpout)
return 0
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
outfile = ""
idListFile = ""
idList = []
infile = ""
i = 1
isNonOptionArg=False
while i < numArgv:
if isNonOptionArg == True:
strs = argv[i].split()
if len(strs) == 1:
idList.append(argv[i])
else:
idList.append(tuple(strs))
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"] :
(idListFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-m", "--m", "-methodid", "--methodid"] :
(g_params['method_getid'], i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-se,", "--se", "-selfield", "--selfield"] :
(tmpint, i) = myfunc.my_getopt_int(argv, i)
g_params['sel_field_list'].append(tmpint)
g_params['method_getid'] = 3
elif argv[i] in ["-i", "--i"] :
(infile, 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:
strs = argv[i].split()
if len(strs) == 1:
idList.append(argv[i])
else:
idList.append(tuple(strs))
i += 1
if infile == "":
print >> sys.stderr, "infile not set"
print >> sys.stderr, usage_short
return 1
elif not os.path.exists(infile):
print >> sys.stderr, "infile %s does not exist"%(infile)
return 1
if idListFile != "":
hdl = myfunc.ReadLineByBlock(idListFile)
if not hdl.failure:
lines = hdl.readlines()
while lines != None:
for line in lines:
if line:
strs = line.split()
if len(strs) == 1:
idList.append(strs[0])
else:
idList.append(tuple(strs))
lines = hdl.readlines()
hdl.close()
if len(idList) < 1:
print >> sys.stderr, "ID not set"
print >> sys.stderr, usage_short
idListSet = set(idList)
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
status = SelectLineByID(infile, idListSet, fpout)
myfunc.myclose(fpout)
return status
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
def main():#{{{
argv = sys.argv
numArgv=len(sys.argv)
if numArgv < 2:
PrintHelp()
sys.exit()
max_numpair = 10*1000*1000
isQuiet = False
rand_seed = None
idList = []
idListFile = ""
outfile=""
method = 0
i = 1
isNonOptionArg=False
while i < numArgv:
if isNonOptionArg == True:
idList.append(argv[i])
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()
sys.exit()
elif sys.argv[i] in [ "-o" , "--o", "-outfile" , "--outfile"]:
outfile, i = myfunc.my_getopt_str(argv,i)
elif argv[i] in ["-m", "--m", "-method", "--method"]:
method, i = myfunc.my_getopt_int(argv,i)
elif sys.argv[i] in [ "-l" , "--l", "-listfile" , "--listfile"]:
idListFile, i = myfunc.my_getopt_str(argv,i)
elif sys.argv[i] in [ "-maxpair" , "--maxpair"]:
max_numpair, i = myfunc.my_getopt_int(argv,i)
elif sys.argv[i] in [ "-seed" , "--seed"]:
rand_seed, i = myfunc.my_getopt_int(argv,i)
elif sys.argv[i] == "-q":
isQuiet=True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", sys.argv[i]
return 1
else:
idList.append(argv[i])
i += 1
if idListFile != "":
idList += myfunc.ReadIDList(idListFile)
numseqid = len(idList)
if numseqid <= 0:
print >> sys.stderr, "List file is empty."
return 1
elif numseqid < 2:
print >> sys.stderr, "Too few items. At least 2 are required."
return 1
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
if method == 0:
pairlist = myfunc.GenerateRandomPair(len(idList), max_numpair,
rand_seed)
elif method == 1:
pairlist = myfunc.GenerateRandomPair_no_repeat_use(len(idList),
max_numpair, rand_seed)
for pair in pairlist:
print >> fpout, "%s %s" %(idList[pair[0]], idList[pair[1]])
myfunc.myclose(fpout)
return 0
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outpath = ""
pairListFile = ""
seqlenFile = ""
shortid2fullidFile = ""
seqid2pfamidMapFile = ""
pfamDefFile = '/data3/data/pfam/pfam27.0/Pfam-A.clans.tsv'
topodb = ""
seqdb = ""
pdb2spFile = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
isNonOptionArg = False
i += 1
return 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 ["-topodb", "--topodb"]:
(topodb, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-pdb2sp", "-pdb2sp", "-pdbtosp", "--pdbtosp"]:
(pdb2spFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-seqdb", "--seqdb"]:
(seqdb, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-seqmsapath", "--seqmsapath"]:
(g_params['seqmsapath'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-datapath", "--datapath"]:
(g_params['datapath'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-seq2pfam", "--seq2pfam"]:
(seqid2pfamidMapFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-pfam2seq", "--pfam2seq"]:
(pfamid2seqidMapFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-description", "--description"]:
(g_params['description'], 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 ["-alignrange", "--alignrange"]:
g_params['alignrange'], i = myfunc.my_getopt_str(argv, i)
if not g_params['alignrange'] in ['all', 'full', 'part']:
print >> sys.stderr, "alignrange must be one of [all, full, part]"
return 1
else:
if g_params['alignrange'] == 'full':
g_params['alignrange'] = 'FULL_ALIGNED'
elif g_params['alignrange'] == 'part':
g_params['alignrange'] = 'PART_ALIGNED'
elif argv[i] in ["-basename", "--basename"]:
(g_params['basename'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-treepath", "--treepath"]:
(g_params['treepath'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-pairalnpath", "--pairalnpath"]:
(g_params['pairalnpath'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-maxperfamily", "--maxperfamily"]:
(g_params['max_num_output_per_family'],
i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-min-seqidt", "--min-seqidt"]:
g_params['minSeqIDT'], i = myfunc.my_getopt_float(argv, i)
elif argv[i] in ["-max-seqidt", "--max-seqidt"]:
g_params['maxSeqIDT'], i = myfunc.my_getopt_float(argv, i)
elif argv[i] in ["-shortid2fullid", "--shortid2fullid"]:
(shortid2fullidFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-debug", "--debug"]:
if argv[i + 1][0].lower() == 'y':
g_params['isDEBUG'] = True
else:
g_params['isDEBUG'] = False
i += 2
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 g_params['basename'] == "":
print >> sys.stderr, "basename not set. exit"
return 1
if myfunc.checkfile(g_params['datapath'], "datapath") != 0:
return 1
if myfunc.checkfile(seqid2pfamidMapFile, "seqid2pfamidMapFile") != 0:
return 1
if myfunc.checkfile(pfamid2seqidMapFile, "pfamid2seqidMapFile") != 0:
return 1
if myfunc.checkfile(topodb + "0.db", "topodb") != 0:
return 1
if myfunc.checkfile(seqdb + "0.db", "seqdb") != 0:
return 1
if myfunc.checkfile(g_params['seqmsapath'], "seqmsapath") != 0:
return 1
if pdb2spFile != "":
(g_params['pdb2uniprotMap'],
g_params['uniprot2pdbMap']) = myfunc.ReadPDBTOSP(pdb2spFile)
if g_params['datapath'] == "":
print >> sys.stderr, "datapath not set"
return 1
elif not os.path.exists(g_params['datapath']):
print >> sys.stderr, "datapath %s does not exist" % (
g_params['datapath'])
return 1
if outpath == "":
print >> sys.stderr, "outpath not set"
return 1
elif not os.path.exists(outpath):
cmd = ["mkdir", "-p", outpath]
subprocess.check_call(cmd)
paircmpfile = "%s/%s.paircmp" % (g_params['datapath'],
g_params['basename'])
if myfunc.checkfile(paircmpfile, "paircmpfile") != 0:
return 1
(g_params['pfamidDefDict'],
g_params['clanidDefDict']) = lcmp.ReadPfamDefFile(pfamDefFile)
g_params['seqid2pfamidDict'] = myfunc.ReadFam2SeqidMap(seqid2pfamidMapFile)
g_params['pfamid2seqidDict'] = myfunc.ReadFam2SeqidMap(pfamid2seqidMapFile)
tmpdir = tempfile.mkdtemp()
if g_params['msapath'] == "":
g_params['msapath'] = tmpdir
if g_params['treepath'] == "":
g_params['treepath'] = tmpdir
if g_params['pairalnpath'] == "":
g_params['pairalnpath'] = tmpdir
pairCmpRecordList = []
unprocessedBuffer = ""
cntTotalReadInRecord = 0
cntTotalOutputRecord = 0
isEOFreached = False
try:
fpin = open(paircmpfile, "r")
except IOError:
print >> sys.stderr, "Failed to open input file %s" % (paircmpfile)
return 1
while 1:
buff = fpin.read(myfunc.BLOCK_SIZE)
if buff == "":
isEOFreached = True
buff = unprocessedBuffer + buff
rdList = []
unprocessedBuffer = lcmp.ReadPairCmpResultFromBuffer(buff, rdList)
rdList = FilterPairCmpResult(rdList)
cntTotalReadInRecord += len(rdList)
pairCmpRecordList += rdList
if isEOFreached == True:
break
fpin.close()
print "cntTotalReadInRecord =", cntTotalReadInRecord
g_params['hdl_seqdb'] = myfunc.MyDB(seqdb)
g_params['hdl_topodb'] = myfunc.MyDB(topodb)
g_params['OS'] = os.uname()[0]
if g_params['OS'].find('Linux') != -1:
g_params['CP_EXE'] = "/bin/cp -uf"
else:
g_params['CP_EXE'] = "/bin/cp -f"
if shortid2fullidFile != "":
g_params['uniprotAC2FullSeqIDMap'] = myfunc.ReadID2IDMap(
shortid2fullidFile)
addname = ""
if g_params['alignrange'] != 'all':
addname += ".%s" % (g_params['alignrange'])
dataTable = {}
# structure of dataTable
# dataTable[pfamid] = {'set_seqid':set(), 'difftopopair':[{'INV':[(id1,id2)]},{'TM2GAP':},{}}
# first read in pairCmpRecordList
AddAllSeqInPairCmp(dataTable, pairCmpRecordList,
g_params['seqid2pfamidDict'])
pairInfoFileList = []
for cmpclass in g_params['cmpClassList_mp3_cmpdup'][0:]:
ss = "%s/%s_.cmpdup.FULL_ALIGNED.%s.pairinfo.txt" % (
g_params['datapath'], g_params['basename'], cmpclass)
pairInfoFileList.append(ss)
pairinfoList = ReadPairInfo_cmpclass(ss)
AddPairInfo(dataTable, pairinfoList, cmpclass)
# print "\n".join(pairInfoFileList)
if g_params['isDEBUG']: #{{{
for pfamid in dataTable:
print pfamid
print "\tset_seqid"
print dataTable[pfamid]['set_seqid']
print "\tdifftopopair"
for cls in dataTable[pfamid]['difftopopair']:
print "\t\t", cls
for tup in dataTable[pfamid]['difftopopair'][cls]:
print "\t\t\t", tup #}}}
WriteHTML(dataTable, outpath)
os.system("rm -rf %s" % (tmpdir))
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
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outpath = "./"
idListFile = None
idList = []
seqfile = ""
topofile = ""
max_dist = 12 # maximal distance to the TM helix so that K, R residues are counted
flank_win = 5 # flanking window of the TM helix, residues at position
#TMbeg-flank_win and TMend+flank_win are also counted
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 ["-outpath", "--outpath"]:
(outpath, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-maxdist", "--maxdist"]:
(max_dist, i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-flankwin", "--flankwin"]:
(flank_win, i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-seqfile", "--seqfile"]:
(seqfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-topofile", "--topofile"]:
(topofile, 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 ["-q"]:
g_params['isQuiet'] = True
i += 1
elif argv[i] in ["-debug"]:
g_params['isDEBUG'] = True
i += 1
else:
print("Error! Wrong argument:", argv[i], file=sys.stderr)
return 1
else:
idList.append(argv[i])
i += 1
(idListSeq, annoListSeq, seqList) = myfunc.ReadFasta(seqfile)
(idListTopo, annoListTopo, topoList) = myfunc.ReadFasta(topofile)
numSeq = len(idListSeq)
numTopo = len(idListTopo)
if numSeq < 1 or numTopo < 1:
print("No seq set", file=sys.stderr)
return 1
seqDict = {}
for i in range(numSeq):
seqDict[idListSeq[i]] = seqList[i]
topoDict = {}
for i in range(numTopo):
topoDict[idListTopo[i]] = topoList[i]
cmpclassDict = {}
for anno in annoListTopo:
anno = anno.lstrip(">")
strs = anno.split()
cmpclassDict[strs[0]] = strs[1]
outpath = os.path.dirname(seqfile)
if outpath == "":
outpath = "."
rootname = os.path.basename(os.path.splitext(seqfile)[0])
outfile_kr_list = outpath + os.sep + rootname + ".krlist.txt"
outfile_krbias = outpath + os.sep + rootname + ".krbias.txt"
fpout_krlist = open(outfile_kr_list, "w")
fpout_krbias = open(outfile_krbias, "w")
for idd in idListSeq:
if g_params['isDEBUG']:
print("seqid: %s" % (idd))
try:
topo = topoDict[idd]
except KeyError:
print("no topo for %s" % idd, file=sys.stderr)
continue
try:
seq = seqDict[idd]
except KeyError:
print("no seq for %s" % idd, file=sys.stderr)
continue
try:
cmpclass = cmpclassDict[idd]
except KeyError:
cmpclass = "INV"
(kr_bias, KR_pos_list, numTM) = CalKRBias(seq, topo, flank_win,
max_dist)
WriteResult(idd, cmpclass, seq, numTM, kr_bias, KR_pos_list,
fpout_krlist)
if cmpclass in ["IDT", "INV"]:
fpout_krbias.write("%d\n" % kr_bias)
fpout_krlist.close()
fpout_krbias.close()
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 = ""
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
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
rmsg = ""
outpath = ""
jobid = ""
datapath = ""
numseq = -1
numseq_this_user = -1
email = ""
host_ip = ""
base_www_url = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
print >> g_params['fperr'], "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 ["-outpath", "--outpath"]:
(outpath, 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 ["-host", "--host"]:
(host_ip, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-nseq", "--nseq"]:
(numseq, i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-nseq-this-user", "--nseq-this-user"]:
(numseq_this_user, i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-baseurl", "--baseurl"]:
(base_www_url, 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 ["-datapath", "--datapath"]:
(datapath, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-force", "--force"]:
g_params['isForceRun'] = True
i += 1
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> g_params['fperr'], "Error! Wrong argument:", argv[i]
return 1
else:
print >> g_params['fperr'], "Error! Wrong argument:", argv[i]
return 1
if outpath == "":
print >> g_params['fperr'], "outpath not set. exit"
return 1
elif not os.path.exists(outpath):
cmd = ["mkdir", "-p", outpath]
try:
rmsg = subprocess.check_output(cmd, stderr=subprocess.STDOUT)
except subprocess.CalledProcessError, e:
print e
print rmsg
return 1
def main(g_params):#{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
fileListFile = ""
fileList = []
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 ["-nmax", "--nmax"]:
(g_params['nmax'], i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-l", "--l"] :
(fileListFile, 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, "%s: no input file is set. exit"%(sys.argv[0])
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
pfamidset_all = set([])
pfamidset_output = set([])
nmax = g_params['nmax']
cnt_round = 0
while 1:
cnt_round += 1
famid2seqidDict = {}
for i in xrange(len(fileList)):
hdl = myfunc.ReadLineByBlock(fileList[i])
if hdl.failure:
continue
lines = hdl.readlines()
while lines != None:
for line in lines:
line = line.strip()
if not line or line[0] == "#":
continue
strs = line.split()
if len(strs) > 2:
seqid = strs[0]
pfamidlist = strs[2:]
for pfamid in pfamidlist:
if cnt_round == 1:
pfamidset_all.add(pfamid)
if pfamid in pfamidset_output:
continue
if not pfamid in famid2seqidDict:
if len(famid2seqidDict) < nmax:
famid2seqidDict[pfamid] = []
if pfamid in famid2seqidDict:
famid2seqidDict[pfamid].append(seqid)
else:
msg="broken item in file %s: line \"%s\""
print >> sys.stderr, msg%(fileList[i], line)
lines = hdl.readlines()
hdl.close()
for pfamid in famid2seqidDict:
pfamidset_output.add(pfamid)
seqidlist = famid2seqidDict[pfamid]
seqidlist = myfunc.uniquelist(seqidlist)
fpout.write("%s %d"%(pfamid, len(seqidlist)))
for seqid in seqidlist:
fpout.write(" %s"%(seqid))
fpout.write("\n")
if len(pfamidset_output) == len(pfamidset_all):
break
else:
print " %d / %d "%(len(pfamidset_output), len(pfamidset_all))
myfunc.myclose(fpout)
if outfile != "":
print "result output to %s"%(outfile)
return 0
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
msapath = ""
outname = ""
pairlistwithfamid_file = ""
isLocalAlignment = False
idListFile = None
idList = []
msaext = ".msa.fa"
i = 1
isNonOptionArg=False
while i < numArgv:
if isNonOptionArg == True:
pairlistwithfamid_file = 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 ["-outname", "--outname"]:
(outname, i) = myfunc.my_getopt_str(argv,i)
elif argv[i] in ["-msapath", "--msapath"]:
(msapath,i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-msaext", "--msaext"]:
(msaext,i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-localali", "--localali"]:
isLocalAlignment = True; i += 1
elif argv[i] in ["-all", "--all"]:
g_params['isOutputAll'] = True; i += 1
elif argv[i] in ["-seqidttype", "--seqidttype"]:
(g_params['seqidttype'],i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-maxsel", "--maxsel"]:
(g_params['maxsel'],i) = myfunc.my_getopt_int(argv,i)
elif argv[i] in ["-verbose", "--verbose"]:
(g_params['verbose'],i) = myfunc.my_getopt_int(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:
pairlistwithfamid_file = argv[i]
i += 1
if pairlistwithfamid_file == "":
print >> sys.stderr, "pairlistwithfamid_file not set"
return 1
if outname == "":
addname = ""
if isLocalAlignment is True:
addname = ".local"
outname = os.path.splitext(pairlistwithfamid_file)[0] + addname + ".selected"
verbose = g_params['verbose']
if verbose >= 1:
print "Reading file", pairlistwithfamid_file
# pairlistDict {pfamid:[(id1,id2),(id1,id2),...] }
pairlistDict = ReadPairListWithFamID(pairlistwithfamid_file)
#print pairlistDict
# obtain pairwise alignment factor
if verbose >= 1:
print "Obtaining pairwise alignment factor"
AddPairwiseAlignmentFactor(pairlistDict, msapath, msaext, isLocalAlignment)
if not g_params['isOutputAll']:
if verbose >=1:
print "Select pair list to limit to ", g_params['maxsel']
pairlistDict = RandSelectPairList(pairlistDict)
#print "selected pairlistDict = ", pairlistDict
#print "pairlistDict_selected=",pairlistDict
if verbose >= 1:
print "Write out the alignment"
WritePairAln(pairlistDict, msapath, msaext, outname)