def FilterSignalPeptide(topofile, sigpepDict, outfile,
isDeleteSeqWithSignalPeptide):
hdl = myfunc.ReadFastaByBlock(topofile)
if hdl.failure:
return 1
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
recordList = hdl.readseq()
while recordList != None:
for rd in recordList:
try:
sp_pos = sigpepDict[rd.seqid]
except KeyError:
sp_pos = -1
if sp_pos != -1:
if isDeleteSeqWithSignalPeptide:
newtopo = ""
else:
newtopo = lcmp.FilterSignalPeptideInTopology(rd.seq, sp_pos)
else:
newtopo = rd.seq
if newtopo != "" and myfunc.CountTM(newtopo) > 0:
fpout.write(">%s\n"%(rd.description))
fpout.write("%s\n"%(newtopo))
recordList = hdl.readseq()
hdl.close()
myfunc.myclose(fpout)
return 0
def MSA2Seq_fasta(infile, outfile): #{{{
try:
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
fpin = open(infile, "rb")
unprocessedBuffer = ""
isEOFreached = False
while 1:
buff = fpin.read(BLOCK_SIZE)
if len(buff) < BLOCK_SIZE:
isEOFreached = True
buff = unprocessedBuffer + buff
recordList = []
unprocessedBuffer = myfunc.ReadFastaFromBuffer(
buff, recordList, isEOFreached)
for rd in recordList:
anno = rd[1]
seq = rd[2].replace("-", "").replace(".", "").replace(" ", "")
fpout.write(">%s\n" % anno)
fpout.write("%s\n" % seq)
if isEOFreached == True:
break
fpin.close()
myfunc.myclose(fpout)
except IOError:
print >> sys.stderr, "Failed to read file", infile
return 1
def MatchMSATopo_using_topofile(
msafile,
topofile,
isIgnoreBadseq, #{{{
method_match,
outfile):
topoDict = GetTopoDict(topofile)
hdl = myfunc.ReadFastaByBlock(msafile)
if hdl.failure:
return 1
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
recordList = hdl.readseq()
while recordList != None:
for rd in recordList:
try:
topo = topoDict[rd.seqid]
except KeyError:
print("topo not found for ID %s" % (rd.seqid), file=sys.stderr)
topo = ""
matchedtopo = MatchSeqToTopo(rd.seq, topo, method_match)
if not (matchedtopo == "BADSEQ" and isIgnoreBadseq):
print(">%s" % (rd.description), file=fpout)
print("%s" % (matchedtopo), file=fpout)
recordList = hdl.readseq()
myfunc.myclose(fpout)
hdl.close()
return 0
def Stat2(lines, outfile):
MAX_NUMTM = 14
dt = {}
for item in [1, 12, 2]:
dt[item] = InitXY(MAX_NUMTM)
cnt = 0
for line in lines:
if line:
strs = line.split()
if strs[0] == "TMMap":
cnt += 1
numTM = int(strs[4].rstrip(":"))
if numTM <= 1 or numTM >= MAX_NUMTM:
continue
mapArray = [int(x) for x in strs[5:]]
st = 1
str_maparray_list = ["%d" % x for x in mapArray]
str_maparray = "".join(str_maparray_list)
num_1 = str_maparray.count("1")
num_2 = str_maparray.count("2")
num_12 = num_1 + num_2
dt[1][1][num_1] += 1
dt[2][1][num_2] += 1
dt[12][1][num_12] += 1
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
fpout.write("#%2s %8s %3s %8s %3s %8s\n" %
("1x", "1y", "2x", "2y", "12x", "12y"))
for i in xrange(1, MAX_NUMTM + 1):
for item in [1, 2, 12]:
(lx, ly) = dt[item]
fpout.write("%3d %8d " % (lx[i], ly[i]))
fpout.write("\n")
myfunc.myclose(fpout)
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
topofile = ""
outfile = ""
isGapLess = False
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
topofile = 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 = argv[i + 1]
i += 2
elif argv[i] in ["-i", "--i"]:
topofile = argv[i + 1]
i += 2
elif argv[i] in ["-gapless", "--gapless"]:
isGapLess = True
i += 1
elif argv[i] in ["-q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
topofile = argv[i]
i += 1
if topofile == "":
print >> sys.stderr, "topofile not set. exit"
return 1
try:
(idList, annoList, seqList) = myfunc.ReadFasta(topofile)
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
for i in xrange(len(idList)):
topo = seqList[i]
seqid = idList[i]
if isGapLess:
topo = topo.replace("-", "").replace(".", "")
posTMList = myfunc.GetTMPosition(topo)
print >> fpout, seqid, posTMList
myfunc.myclose(fpout)
except (IOError, IndexError):
pass
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(g_params):#{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
infile = ""
idwithannofile = ""
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 = argv[i+1]
i += 2
elif argv[i] in ["-case", "--case"]:
g_params['isCaseSensitive'] = True
i += 1
elif argv[i] in ["-idwithanno", "--idwithanno"]:
idwithannofile = argv[i+1]
i += 2
elif argv[i] in ["-key", "--key"]:
g_params['keywordList'].append(argv[i+1])
i += 2
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 infile == "":
print >> sys.stderr, "infile not set.exit"
return 1
if idwithannofile == "":
print >> sys.stderr, "idwithannofile not set.exit"
return 1
seqid2AnnoDict = ReadSeqIDWithAnnoFile(idwithannofile)
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
FilterUniprotTableinfoByKeyword(infile, seqid2AnnoDict, fpout)
myfunc.myclose(fpout)
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
fileListFile = ""
fileList = []
isOverWrite = 0
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 ["-outpath", "--outpath"]:
outpath = argv[i + 1]
i += 2
elif argv[i] in ["-l", "--l"]:
fileListFile = argv[i + 1]
i += 2
elif argv[i] in ["-overwrite", "--overwrite"]:
isOverWrite = 1
i += 1
elif argv[i] in ["-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 != "":
try:
fp = open(fileListFile, "r")
fileList += fp.read().split()
fp.close()
except IOError:
msg = "Failed to read idlistfile {}."
print >> sys.stderr, msg.format(fileListFile)
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
for i in xrange(len(fileList)):
CleanAASeq(fileList[i], isOverWrite, 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 = ""
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): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
idListFile = ""
idList = []
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
idList.append(argv[i])
i += 1
isNonOptionArg = False
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", "-list"]:
(idListFile, i) = myfunc.my_getopt_str(argv, i)
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
idList.append(argv[i])
i += 1
if idListFile != "":
idList += myfunc.ReadIDList(idListFile)
numID = len(idList)
if numID < 1:
print >> sys.stderr, "No ID set. exit"
return 1
params = {}
params['from'] = 'P_GI'
params['to'] = 'ID' # to uniprot id
params['format'] = 'tab'
params['query'] = " ".join(idList)
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
GIID2UniprotID(params, fpout)
myfunc.myclose(fpout)
def OutputPfamFastaFile(seqidList, pfamid, seqdbDict, hdl_seqdb,
extra_desp_dict, outpath): #{{{
outfile = "%s%s%s%s" % (outpath, os.sep, pfamid, g_params['out_ext'])
fpout = myfunc.myopen(outfile, None, "w", True)
isAddExtraDescription = False
if len(extra_desp_dict) > 0:
isAddExtraDescription = True
for seqid in seqidList:
if seqid.find("UniRef") != -1:
try:
ss = seqid.split("_")
seqid = ss[1]
except IndexError:
pass
if g_params['isBigmem']:
try:
record = seqdbDict[seqid]
(tmpanno, tmpseq) = record
if isAddExtraDescription:
try:
extraanno = extra_desp_dict[seqid]
except KeyError:
extraanno = ""
if extraanno != "":
tmpanno = "%s %s" % (extraanno, tmpanno)
fpout.write(">%s\n%s\n" % (tmpanno, tmpseq))
except KeyError:
print >> sys.stderr, "seqid %s not found in seqdb" % (seqid)
else:
record = hdl_seqdb.GetRecord(seqid)
if record:
if isAddExtraDescription:
try:
extraanno = extra_desp_dict[seqid]
except KeyError:
extraanno = ""
if extraanno == "":
fpout.write("%s" % (record))
else:
(tmpseqid, tmpanno,
tmpseq) = myfunc.ExtractFromSeqWithAnno(record)
tmpanno = "%s %s" % (extraanno, tmpanno)
fpout.write(">%s\n%s\n" % (tmpanno, tmpseq))
else:
fpout.write("%s" % (record))
else:
print >> sys.stderr, "seqid %s not found in seqdb" % (seqid)
myfunc.myclose(fpout)
if g_params['isGzip']:
cmd = ["gzip", "-N", "-f", outfile]
print " ".join(cmd)
subprocess.check_call(cmd, stdout=open(os.devnull, "w"))
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(g_params):#{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outpath = "./"
outfile = ""
fileListFile = ""
fileList = []
pfamDefFile = "%s/data/pfam/pfam26.0/Pfam-A.clans.tsv"%(DATADIR3)
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 ["-outpath", "--outpath"]:
(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 ["-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)
(pfamidDefDict, clanidDefDict) = ReadPfamDefFile(pfamDefFile)
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
for i in xrange(len(fileList)):
CountUniquePairInvertedInfo(fileList[i], pfamidDefDict, fpout)
myfunc.myclose(fpout)
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
allinfoFile = ""
idwithannofile = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
allinfoFile = 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 = argv[i + 1]
i += 2
elif argv[i] in ["-idwithanno", "--idwithanno"]:
idwithannofile = argv[i + 1]
i += 2
elif argv[i] in ["-q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
allinfoFile = argv[i]
i += 1
if allinfoFile == "":
print >> sys.stderr, "allinfoFile not set.exit"
return 1
if idwithannofile == "":
print >> sys.stderr, "idwithannofile not set.exit"
return 1
seqid2AnnoDict = ReadSeqIDWithAnnoFile(idwithannofile)
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
FilterTopconsSingleResultUniprot(allinfoFile, seqid2AnnoDict, fpout)
myfunc.myclose(fpout)
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
clstrfile = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
clstrfile = 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 = argv[i + 1]
i += 2
elif argv[i] in ["-q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
clstrfile = argv[i]
i += 1
if clstrfile == "" or not os.path.exists(clstrfile):
print >> sys.stderr, "clstrfile file not set or not exist"
return 1
fpout = sys.stdout
if outfile != "":
try:
fpout = open(outfile, "w")
except IOError:
print >> sys.stderr, "Failed to write to file %s" % outfile
fpout = sys.stdout
ClstrToPairList(clstrfile, 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("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 ReWriteFasta(infile, outfile): #{{{
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
hdl = myfunc.ReadFastaByBlock(infile, 0, 1)
if hdl.failure:
return 1
recordList = hdl.readseq()
while recordList != None:
for rd in recordList:
fpout.write(">%s\n" % rd.description)
fpout.write("%s\n" % rd.seq)
recordList = hdl.readseq()
hdl.close()
myfunc.myclose(fpout)
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():#{{{
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
infile = ""
outfile = ""
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"]:
outfile = argv[i + 1]
i += 2
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 infile == "":
print >> sys.stderr, "infile not set. exit"
return 1
elif not os.path.exists(infile):
print >> sys.stderr, "infile %s does not exist. exit" % (infile)
return 1
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
FilterUniprotIDMap(infile, fpout)
myfunc.myclose(fpout)
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 GetSeqFromMSA(infile, outfile, hdl_seqdb):#{{{
if not os.path.exists(infile):
print >> sys.stderr, "infile %s does not exist."%(infile)
return 1
outdir = os.path.dirname(outfile)
if outdir != "" and not os.path.exists(outdir):
os.system("mkdir -p %s"%(outdir))
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
filetype = GetFileType(infile)
inputfile = ""
if filetype == "fasta":
tmpf = tempfile.mktemp()
os.system("grep '^>' %s | sed 's/>//' > %s"%(infile, tmpf))
inputfile = tmpf
try:
first_record = SeqIO.parse(open(infile, "rU"), "fasta").next()
if first_record.id == "target":
seq = (first_record.seq._data).replace("-","")
fpout.write(">%s\n"%(first_record.description))
fpout.write("%s\n"%(seq))
except (IOError,ValueError,KeyError):
pass
elif filetype == "idlist":
inputfile = infile
else:
print >> sys.stderr, "Unrecognized infile type"
return 1
(status, numseq, numRetrieved) = GetFullSeq(inputfile, hdl_seqdb, fpout)
if status == 1:
msg = "%s retrieved %d out of %d sequences. Failed to read."
else:
msg = "%s retrieved %d out of %d sequences. Succeeded."
print msg%(infile, numRetrieved, numseq)
if filetype == "fasta":
os.system("rm -f %s"%(tmpf))
myfunc.myclose(fpout)
return 0
def Stat1(lines, outfile):
MAX_NUMTM = 14
dt = {}
for item in [0.0, 0.5, 1.0]:
dt[item] = InitXY(MAX_NUMTM)
cnt = 0
for line in lines:
if line:
strs = line.split()
if strs[0] == "TMMap":
cnt += 1
numTM = int(strs[4].rstrip(":"))
if numTM <= 1 or numTM >= MAX_NUMTM:
continue
mapArray = [int(x) for x in strs[5:]]
st = 1
str_maparray_list = ["%d" % x for x in mapArray]
str_maparray = "".join(str_maparray_list)
posContList = GetSegPos(str_maparray, "%d" % st)
neighbour_char = "0"
posContList = FilterSegPos(posContList, str_maparray,
neighbour_char)
if len(posContList) >= 1:
for (b, e) in posContList:
if b == 0:
pp = 0.0
elif e == numTM:
pp = 1.0
else:
pp = 0.5
if DEBUG and (e - b) == 1:
print "Maparray", mapArray, posContList
dt[pp][1][e - b] += 1
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
fpout.write("#%2s %8s %3s %8s %3s %8s\n" %
("Nx", "Ny", "Ix", "Iy", "Cx", "Cy"))
for i in xrange(1, MAX_NUMTM + 1):
for item in [0.0, 0.5, 1.0]:
(lx, ly) = dt[item]
fpout.write("%3d %8d " % (lx[i], ly[i]))
fpout.write("\n")
myfunc.myclose(fpout)
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)
IDMap2SeqID(infile, fpout)
myfunc.myclose(fpout)
def RemoveDupSeq(infile, g_outpath, method, isUseMD5):#{{{
if g_outpath == "":
outpath = myfunc.my_dirname(infile)
else:
outpath = g_outpath
rootname = os.path.basename(os.path.splitext(infile)[0])
outfile = "%s%s%s"%(outpath, os.sep, rootname)
fpout = myfunc.myopen(outfile, None, "w", False)
if fpout == None:
return 1
hdl = myfunc.ReadFastaByBlock(infile)
if hdl.failure:
return -1
myset = set([])
recordList = hdl.readseq()
while recordList != None:
for rd in recordList:
if method == "id":
key = rd.seqid
elif method == "seq":
if isUseMD5:
key = md5.new(rd.seq).digest()
else:
key = rd.seq
if not key in myset:
myset.add(key)
fpout.write(">%s\n%s\n"%(rd.description, rd.seq))
recordList = hdl.readseq()
hdl.close()
myfunc.myclose(fpout)
return 0
def MatchMSATopo_using_topodb(
msafile,
topodb,
isIgnoreBadseq, #{{{
method_match,
outfile):
hdl_topo = myfunc.MyDB(topodb)
if hdl_topo.failure:
return 1
hdl = myfunc.ReadFastaByBlock(msafile)
if hdl.failure:
return 1
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
recordList = hdl.readseq()
while recordList != None:
for rd in recordList:
topowithanno = hdl_topo.GetRecord(rd.seqid)
if topowithanno != None:
(topoid, topoanno,
topo) = myfunc.ExtractFromSeqWithAnno(topowithanno)
else:
print("topo not found for ID %s" % (rd.seqid), file=sys.stderr)
topo = ""
matchedtopo = MatchSeqToTopo(rd.seq, topo, method_match)
if not (matchedtopo == "BADSEQ" and isIgnoreBadseq):
print(">%s" % (rd.description), file=fpout)
print("%s" % (matchedtopo), file=fpout)
recordList = hdl.readseq()
myfunc.myclose(fpout)
hdl.close()
hdl_topo.close()
return 0
def RemoveDupSeq(infile, g_outpath, method, isUseMD5): #{{{
if g_outpath == "":
outpath = myfunc.my_dirname(infile)
else:
outpath = g_outpath
rootname = os.path.basename(os.path.splitext(infile)[0])
outfile = "%s%s%s" % (outpath, os.sep, rootname)
fpout = myfunc.myopen(outfile, None, "w", False)
if fpout == None:
return 1
hdl = myfunc.ReadFastaByBlock(infile)
if hdl.failure:
return -1
myset = set([])
recordList = hdl.readseq()
while recordList != None:
for rd in recordList:
if method == "id":
key = rd.seqid
elif method == "seq":
if isUseMD5:
key = md5.new(rd.seq).digest()
else:
key = rd.seq
if not key in myset:
myset.add(key)
fpout.write(">%s\n%s\n" % (rd.description, rd.seq))
recordList = hdl.readseq()
hdl.close()
myfunc.myclose(fpout)
return 0
def main(g_params):
numArgv=len(sys.argv)
if numArgv < 2:
PrintHelp()
return 1
outFile = ""
orderlistfile = ""
msafile = ""
outformat = "fasta" # fasta or anno
i = 1
isNonOptionArg=False
while i < numArgv:
if isNonOptionArg == True:
msafile = 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] == "-h" or sys.argv[i] == "--help":
PrintHelp()
return 1
elif sys.argv[i] in [ "-o", "--o"] :
outFile=sys.argv[i+1]
i = i + 2
elif sys.argv[i] == "-orderlist" or sys.argv[i] == "--orderlist":
orderlistfile = sys.argv[i+1]
i = i + 2
elif sys.argv[i] == "-msafile" or sys.argv[i] == "--msafile":
msafile = sys.argv[i+1]
i = i + 2
elif sys.argv[i] in ["-of", "--of", "-outformat", "--outformat"]:
outformat = sys.argv[i+1].lower()
i += 2
else:
print(("Error! Wrong argument:%s" % sys.argv[i]), file=sys.stderr)
return 1
else:
msafile = sys.argv[i]
i+=1
if not outformat in ["anno", "fasta"]:
print("Unrecognized outformat \"%s\","%(
outformat) + " should be either \"anno\" or \"fasta\".", file=sys.stderr)
return 1
if orderlistfile == "":
print("orderlist file not set. Exit", file=sys.stderr)
return 1
if msafile == "":
print("msafile not set. Exit", file=sys.stderr)
orderList = ReadOrderList(orderlistfile)
(idList, annoList, seqList) = myfunc.ReadFasta(msafile)
if len(orderList) > 0 and len(idList) > 0:
fpout = sys.stdout
fpout = myfunc.myopen(outFile, sys.stdout, "w", False)
seqDict = {}
annoDict = {}
numSeq = len(idList)
for i in range(numSeq):
annoDict[idList[i]] = annoList[i]
if outformat != "anno":
for i in range(numSeq):
seqDict[idList[i]] = seqList[i]
for sid in orderList:
if sid in annoDict:
fpout.write(">%s\n"%annoDict[sid])
if outformat != "anno":
fpout.write("%s\n"%seqDict[sid])
else:
print("seqid %s not in msafile %s"%(
sid, msafile), file=sys.stderr)
myfunc.myclose(fpout)
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
outfile = ""
seqdbfile = ""
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 ["-outfile", "--outfile"]:
outfile = argv[i + 1]
i += 2
elif argv[i] in ["-seqdb", "--seqdb"]:
seqdbfile = argv[i + 1]
i += 2
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 infile == "":
print >> sys.stderr, "annotation file not set"
return 1
elif not os.path.exists(infile):
print >> sys.stderr, "annotation file %s does not exist" % (infile)
return 1
if seqdbfile == "":
print >> sys.stderr, "seqdbfile file not set"
return 1
elif not os.path.exists(seqdbfile):
print >> sys.stderr, "seqdbfile file %s does not exist" % (seqdbfile)
return 1
seqDict = GetSeqDict(seqdbfile)
if seqDict == {}:
print >> sys.stderr, "Failed to read seqdbfile %s" % (seqdbfile)
return 1
(idList, annoList, contentList) = myfunc.ReadFasta(infile)
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
for i in xrange(len(idList)):
seqid = idList[i]
try:
seq = seqDict[seqid]
fpout.write(">%s\n" % (annoList[i]))
fpout.write("%s\n" % (seq))
if contentList[i] != "":
fpout.write("%s\n" % (contentList[i]))
except KeyError:
print >> sys.stderr, "seqid %s not found in seqdb" % (seqid)
myfunc.myclose(fpout)
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
datafile = ""
restrictOCList = []
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
datafile = 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 = argv[i + 1]
i += 2
elif argv[i] in ["-i", "--i"]:
datafile = argv[i + 1]
i += 2
elif argv[i] in ["-keep_isoform", "--keep_isoform"]:
g_params['filter_isoform'] = False
i += 1
elif argv[i] in ["-keep_non_refpro", "--keep_non_refpro"]:
g_params['filter_non_refpro'] = False
i += 1
elif argv[i] in ["-keep_no_genename", "--keep_no_genename"]:
g_params['filter_no_genename'] = False
i += 1
elif argv[i] in ["-oc", "--oc"]:
restrictOCList.append(argv[i + 1])
i += 2
elif argv[i] in ["-q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
datafile = argv[i]
i += 1
if not os.path.exists(datafile):
print >> sys.stderr, "datafile %s not set or not exists. Exit" % (
datafile)
return 1
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
UniprotTremblData2Table(datafile, restrictOCList, fpout)
myfunc.myclose(fpout)
return 0
try:
outfile = sys.argv[3]
except:
pass
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
hdl = myfunc.ReadLineByBlock(jobruntimeFile)
if hdl.failure:
sys.exit(1)
lines = hdl.readlines()
cnt=0
while lines != None:
for line in lines:
strs = line.split("\t")
if len(strs) < 8:
continue
jobid = strs[0]
idx = strs[1]
cnt += 1
OutputRecord(cnt, jobid, idx, path_result, fpout)
lines = hdl.readlines()
hdl.close()
myfunc.myclose(fpout)
def CalculateQueue(resultdir, outfile): # {{{
# 1. get the list of working folders
raw_folder_list = os.listdir(resultdir)
folder_nr_list = []
for folder in raw_folder_list:
if os.path.isdir(resultdir + "/" + folder) and (
folder.isdigit() or folder[0:2] == "r_" or folder[:4] == "rst_"
):
folder_nr_list.append(folder)
# 2. gather information for queued jobs
job_table_in_queue = {}
other_job_table = {}
freq_user_in_queue = {} # count the frequency of the user of jobs in queue
freq_user_running = {} # count the frequency of the user for running jobs
for folder in folder_nr_list:
workdir = "%s/%s" % (resultdir, folder)
status = get_job_status(workdir)
if status in ["Queued"]: # jobs in queue
email = ReadContent("%s/%s" % (workdir, "email"))
host = ReadContent("%s/%s" % (workdir, "host"))
date_str = ReadContent("%s/%s" % (workdir, "date"))
sequence = ReadContent("%s/%s" % (workdir, "sequence"))
length_seq = len(sequence)
user = ""
if email and email != "N/A":
user = email
else:
user = host
if not user in freq_user_in_queue:
freq_user_in_queue[user] = 0
freq_user_in_queue[user] += 1
try:
date_submitted = datetime.datetime.strptime(date_str, "%Y-%m-%d %H:%M:%S")
except ValueError:
print >>sys.stderr, "datefile = '%s'. date = '%s'" % ("%s/date" % (workdir), date_str)
print >>sys.stderr, "Ignore %s" % folder
continue
date_now = datetime.datetime.now()
queue_time = date_now - date_submitted
queue_time_in_sec = get_total_seconds(queue_time)
job_table_in_queue[folder] = [status, user, queue_time_in_sec, length_seq]
else:
email = ReadContent("%s/%s" % (workdir, "email"))
host = ReadContent("%s/%s" % (workdir, "host"))
user = ""
if email and email != "N/A":
user = email
else:
user = host
if status in ["Running", "Rerun"]:
if not user in freq_user_running:
freq_user_running[user] = 0
freq_user_running[user] += 1
other_job_table[folder] = [status, user, 0, 0]
else:
other_job_table[folder] = [status, user, 0, 0]
for folder in job_table_in_queue:
user = job_table_in_queue[folder][1]
freq_in_queue = 1
freq_running = 0
if user != "":
freq_in_queue = freq_user_in_queue[user]
try:
freq_running = freq_user_running[user]
except KeyError:
freq_running = 0
job_table_in_queue[folder].append(freq_in_queue)
job_table_in_queue[folder].append(freq_running)
for folder in other_job_table:
other_job_table[folder].append(0)
other_job_table[folder].append(0)
# calculate the priority
# now each job_table_in_queue[folder] has five element
# [user, queue_time_in_sec, length_seq, freq_in_queue, freq_running]
# Group the jobs in each user, and for the sublist of each user, first rank by
# the queue_time_in_sec, and then do as follows
# Note, for those target with <= 100 aa, the run time is relatively similar
# score = queue_time_in_sec / ((rank+addition)**1.5 * max(length_seq,100)**1.5)
# where addition = freq_running
for user in freq_user_in_queue:
sub_table = {}
for folder in job_table_in_queue:
if job_table_in_queue[folder][1] == user:
sub_table[folder] = job_table_in_queue[folder]
# in descending order by queue_time_in_sec
sorted_sub_table = sorted(sub_table.items(), key=lambda x: x[1][2], reverse=True)
for i in xrange(len(sorted_sub_table)):
folder_nr = sorted_sub_table[i][0]
queue_time_in_sec = sorted_sub_table[i][1][2]
length_seq = sorted_sub_table[i][1][3]
freq_in_queue = sorted_sub_table[i][1][4]
freq_running = sorted_sub_table[i][1][5]
rank = i + 1
addition = freq_running
if user == "":
rank = 1
addition = 0
score = queue_time_in_sec / ((rank + addition) ** 2 * max(length_seq, 100) ** 1.5)
job_table_in_queue[folder_nr].append(score)
for folder in other_job_table:
other_job_table[folder].append(0)
# now rank the job_table_in_queue again
sorted_job_table_in_queue = sorted(job_table_in_queue.items(), key=lambda x: x[1][6], reverse=True)
# write the result
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
print >> fpout, "#%-5s %8s %4s %-30s %10s %10s %6s %6s" % (
"ID",
"Status",
"Rank",
"User",
"PD_time(s)",
"Score",
"Count_PD",
"Count_R",
)
for i in xrange(len(sorted_job_table_in_queue)):
folder = sorted_job_table_in_queue[i][0]
rank = i + 1
status = sorted_job_table_in_queue[i][1][0]
user = sorted_job_table_in_queue[i][1][1]
queue_time_in_sec = sorted_job_table_in_queue[i][1][2]
freq_in_queue = sorted_job_table_in_queue[i][1][4]
freq_running = sorted_job_table_in_queue[i][1][5]
score = sorted_job_table_in_queue[i][1][6]
print >> fpout, "%-6s %8s %4d %-30s %10.1f %10.1f %6d %6d" % (
folder,
status,
rank,
user,
queue_time_in_sec,
score,
freq_in_queue,
freq_running,
)
# now rank the job_table_in_queue again
sorted_other_job_table = sorted(other_job_table.items(), key=lambda x: x[1][0]) # sorted by status
for i in xrange(len(sorted_other_job_table)):
folder = sorted_other_job_table[i][0]
rank = 0
status = sorted_other_job_table[i][1][0]
user = sorted_other_job_table[i][1][1]
queue_time_in_sec = sorted_other_job_table[i][1][2]
freq_in_queue = sorted_other_job_table[i][1][4]
freq_running = sorted_other_job_table[i][1][5]
score = sorted_other_job_table[i][1][6]
print >> fpout, "%-6s %8s %4d %-30s %10.1f %10.1f %6d %6d" % (
folder,
status,
rank,
user,
queue_time_in_sec,
score,
freq_in_queue,
freq_running,
)
myfunc.myclose(fpout)