def Ana_NumContUnmappedTM(infile):
methodList = [0,1,2,3]
outpath = os.path.dirname(infile)
if outpath == "":
outpath = "."
try:
freqDict = {}
for method in methodList:
freqDict[2*method] = {}
freqDict[2*method+1] = {}
unprocessedBuffer=""
cntTotalReadInRecord = 0
cntTotalOutputRecord = 0
isEOFreached = False
fpin = open(infile)
while 1:
buff = fpin.read(BLOCK_SIZE)
if buff == "":
isEOFreached = True
buff = unprocessedBuffer + buff
pairCmpRecordList=[]
unprocessedBuffer = lcmp.ReadPairCmpResultFromBuffer(buff,pairCmpRecordList)
if len(pairCmpRecordList) > 0:
CountContinuousUnmappedTM_0(pairCmpRecordList, freqDict[0],
isFilterNeighbour=True)
CountContinuousUnmappedTM_0(pairCmpRecordList, freqDict[1],
isFilterNeighbour=False)
CountContinuousUnmappedTM_1(pairCmpRecordList, freqDict[2],
isFilterNeighbour=True)
CountContinuousUnmappedTM_1(pairCmpRecordList, freqDict[3],
isFilterNeighbour=False)
CountContinuousUnmappedTM_2(pairCmpRecordList, freqDict[4],
isFilterNeighbour=True)
CountContinuousUnmappedTM_2(pairCmpRecordList, freqDict[5],
isFilterNeighbour=False)
CountContinuousUnmappedTM_3(pairCmpRecordList, freqDict[6],
isFilterNeighbour=True)
CountContinuousUnmappedTM_3(pairCmpRecordList, freqDict[7],
isFilterNeighbour=False)
cntTotalReadInRecord += len(pairCmpRecordList)
if isEOFreached == True:
break
fpin.close()
for method in methodList:
for idx in [2*method, 2*method+1]:
if idx == 2*method:
str_filter = "True"
outfile = (outpath + os.sep +
"tmp_ana_numContTM_method%d_filternb.txt" %
(method))
else:
str_filter = "False"
outfile = (outpath + os.sep +
"tmp_ana_numContTM_method%d_nonfilternb.txt" %
(method))
fpout = open(outfile, "w")
print
print >> fpout, "#numTM count Method_%d isFilterNeighbour=%s"%(method,
str_filter)
for i in range(1, 21):
msg = "%-5d %5d"
try:
print >> fpout, msg%(i, freqDict[idx][i])
except KeyError:
print >> fpout, msg%(i, 0)
fpout.close()
cmd = "/data3/wk/MPTopo/src/tmp_plot_histogram_logscale.sh %s"
os.system(cmd%(outfile))
except IOError:
return 1
def main(): #{{{
numArgv = len(sys.argv)
if numArgv < 2:
PrintHelp()
return 1
parameters = {}
parameters['minGapFraction'] = 0.5
parameters['maxGapFraction'] = 1.0
parameters['minDGvalue'] = -999999.0
parameters['maxDGvalue'] = 1.0
infile = ""
outfile = ""
isQuiet = False
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] == "-h" or sys.argv[i] == "--help":
PrintHelp()
sys.exit()
elif (sys.argv[i] == '-o' or sys.argv[i] == '--o'
or sys.argv[i] == "-outfile" or sys.argv[i] == "--outfile"):
outfile = sys.argv[i + 1]
i += 2
elif sys.argv[i] == "-gap" or sys.argv[i] == "--gap":
parameters['minGapFraction'] = float(sys.argv[i + 1])
i += 2
elif sys.argv[i] == "-dg" or sys.argv[i] == "--dg":
parameters['maxDGvalue'] = float(sys.argv[i + 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
rootname = os.path.basename(os.path.splitext(infile)[0])
fpout = sys.stdout
if outfile != "":
try:
fpout = open(outfile, "w")
except IOError:
print >> sys.stderr, "Failed to write to file %s." % outfile
print >> sys.stderr, "Reset output to sys.stdout."
fpout = sys.stdout
pass
fpin = open(infile, "rb")
if not fpin:
print >> sys.stderr, "Failed to open input file %s" % (infile)
return -1
unprocessedBuffer = ""
cntTotalOutputRecord = 0
cntTotalReadInRecord = 0
isEOFreached = False
while 1:
buff = fpin.read(BLOCK_SIZE)
if buff == "":
isEOFreached = True
buff = unprocessedBuffer + buff
pairCmpRecordList = []
unprocessedBuffer = lcmp.ReadPairCmpResultFromBuffer(
buff, pairCmpRecordList)
if len(pairCmpRecordList) > 0:
#WritePairCmpRecord(pairCmpRecordList,fpout);
filteredList = FilterPairCmpResult(pairCmpRecordList, parameters)
(status, cntTotalOutputRecord) = lcmp.WritePairCmpRecord(
filteredList, cntTotalOutputRecord, fpout)
cntTotalReadInRecord += len(pairCmpRecordList)
if isEOFreached == True:
break
fpin.close()
print "cntTotalReadInRecord =", cntTotalReadInRecord
print "cntTotalOutputRecord =", cntTotalOutputRecord
if fpout != None and fpout != sys.stdout:
fpout.close()
return 0
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 Ana_NumTMHeatMap(
infile,
seqid2pfamidDict,
seqid2clanidDict, #{{{
tm_pfamidSet,
tm_clanidSet,
pfamidDefDict,
clanidDefDict,
signalpDict,
classList_TableNumTMHeatMap,
SPE_PAIR_LIST,
pfamid2seqidDict,
clanid2seqidDict,
idSet_TMpro,
usedPfamIDSet,
alignrange):
dataTableNumTMHeatMap = {}
InitTableNumTMHeatMap(dataTableNumTMHeatMap, classList_TableNumTMHeatMap,
100, SPE_PAIR_LIST)
pairInfoListDict = {}
for cls in classList_TableNumTMHeatMap:
pairInfoListDict[cls] = []
if g_params['outpath'] != "":
outpath = g_params['outpath']
else:
outpath = os.path.dirname(infile)
if outpath == "":
outpath = "."
try:
unprocessedBuffer = ""
cntTotalReadInRecord = 0
cntTotalOutputRecord = 0
isEOFreached = False
fpin = open(infile)
while 1:
buff = fpin.read(BLOCK_SIZE)
if buff == "":
isEOFreached = True
buff = unprocessedBuffer + buff
pairCmpRecordList = []
unprocessedBuffer = lcmp.ReadPairCmpResultFromBuffer(
buff, pairCmpRecordList)
if len(pairCmpRecordList) > 0:
#CountSpecialPair(pairCmpRecordList, pairInfoLists, SPE_PAIR_LIST)
AnaPairCmpResultNumTMHeatMap(pairCmpRecordList,
dataTableNumTMHeatMap,
pairInfoListDict,
classList_TableNumTMHeatMap,
signalpDict, SPE_PAIR_LIST,
alignrange)
cntTotalReadInRecord += len(pairCmpRecordList)
print "cntTotalReadInRecord = ", cntTotalReadInRecord
if isEOFreached == True:
break
fpin.close()
#print "count 6,12 = ", dataTableNumTMHeatMap['RMSP']['data'][6][12]
#print "count = ", dataTableNumTMHeatMap['RMSP']['data']
#print "count special", dataTableNumTMHeatMap['RMSP']['pairInfoLists']
try:
for i in xrange(len(
dataTableNumTMHeatMap['RMSP']['pairInfoLists'])):
print SPE_PAIR_LIST[i], len(
dataTableNumTMHeatMap['RMSP']['pairInfoLists'][i])
except KeyError:
pass
# if g_params['numTMHeatMapMode'] == "full":
# FillSymmetricDataTableNumTMHeatMap(dataTableNumTMHeatMap,
# classList_TableNumTMHeatMap)
for cls in classList_TableNumTMHeatMap: # ["ALL", "RMSP"]
(freqListPfam, freqListClan) = AnaFamFrequency_onelist(
pairInfoListDict[cls], seqid2pfamidDict, seqid2clanidDict,
pfamid2seqidDict, clanid2seqidDict, tm_pfamidSet, tm_clanidSet,
idSet_TMpro, usedPfamIDSet)
#print "tm_clanidSet", tm_clanidSet
if g_params['pairwise_comparison_method'] == 1:
cmpclassList = cmpClassList_method1
elif g_params['pairwise_comparison_method'] == 3:
cmpclassList = cmpClassList_method3
isCmpDup = False
outFileFamPairCount = "%s%s%s.%s.%s.pfam.paircount.txt" % (
outpath, os.sep, g_params['outname'], alignrange, cls)
WriteFamPairCount(freqListPfam, pairInfoListDict[cls],
pfamidDefDict, cmpclassList,
g_params['pairwise_comparison_method'], isCmpDup,
outFileFamPairCount)
outFileFamPairCount = "%s%s%s.%s.%s.clan.paircount.txt" % (
outpath, os.sep, g_params['outname'], alignrange, cls)
WriteFamPairCount(freqListClan, pairInfoListDict[cls],
clanidDefDict, cmpclassList,
g_params['pairwise_comparison_method'], isCmpDup,
outFileFamPairCount)
if g_params['pairwise_comparison_method'] == 3:
# if mp=3, write another statistics with cmpdup
isCmpDup = True
cmpclassList = cmpClassList_mp3_cmpdup
outFileFamPairCount = "%s%s%s.%s.%s.cmpdup.pfam.paircount.txt" % (
outpath, os.sep, g_params['outname'], alignrange, cls)
WriteFamPairCount(freqListPfam, pairInfoListDict[cls],
pfamidDefDict, cmpclassList,
g_params['pairwise_comparison_method'],
isCmpDup, outFileFamPairCount)
outFileFamPairCount = "%s%s%s.%s.%s.cmpdup.clan.paircount.txt" % (
outpath, os.sep, g_params['outname'], alignrange, cls)
WriteFamPairCount(freqListClan, pairInfoListDict[cls],
clanidDefDict, cmpclassList,
g_params['pairwise_comparison_method'],
isCmpDup, outFileFamPairCount)
for mode_norm in ["norm_diag", "no_norm"]:
if mode_norm in ["norm_diag", "no_norm"]:
heatmapmode = 'half'
else:
heatmapmode = 'full'
outFileNumTMHeatMap = "%s%s%s.%s.%s.%s.%s.txt" % (
outpath, os.sep, g_params['outname'], alignrange,
heatmapmode, cls, mode_norm)
if heatmapmode == 'full':
mtx = myfunc.FillSymmetricMatrix(
dataTableNumTMHeatMap[cls]['data'],
dataTableNumTMHeatMap[cls]['maxNumTM'])
else:
mtx = dataTableNumTMHeatMap[cls]['data']
if mode_norm == "no_norm":
for i in range(dataTableNumTMHeatMap[cls]['maxNumTM']):
mtx[i][i] = 0
if WriteNumTMHeatMap(mtx,
dataTableNumTMHeatMap[cls]['maxNumTM'],
dataTableNumTMHeatMap[cls]['numPair'],
mode_norm, outFileNumTMHeatMap) == 0:
print "heatmap %s output" % (outFileNumTMHeatMap)
cmd = "%s/plotNumTMHeatMap.sh %s" % (binpath,
outFileNumTMHeatMap)
os.system(cmd)
outFileSpecialPairAna = "%s%s%s.%s.%s.%s.%s.specialpairana.txt" % (
outpath, os.sep, g_params['outname'], alignrange,
g_params['numTMHeatMapMode'], cls, mode_norm)
WriteSpecialPair(dataTableNumTMHeatMap[cls],
pairInfoListDict[cls], seqid2pfamidDict,
seqid2clanidDict, tm_pfamidSet, tm_clanidSet,
pfamidDefDict, clanidDefDict, SPE_PAIR_LIST,
outFileSpecialPairAna)
print "Anafile %s output" % (outFileSpecialPairAna)
# for i in xrange(len(SPE_PAIR_LIST)):
# print
# pair = SPE_PAIR_LIST[i]
# print pair
# print len(pairInfoLists[i])
# print pairInfoLists[i]
except IOError:
return 1
def main():#{{{
numArgv=len(sys.argv)
if numArgv < 2:
PrintHelp()
return 1;
parameters={};
parameters['minGapFraction'] = 0.5;
parameters['maxGapFraction'] = 1.0;
parameters['minDGvalue'] = -999999.0;
parameters['maxDGvalue'] = 1.0;
parameters['minSeqIDT'] = 0.0;
parameters['maxSeqIDT'] = 100.0;
infile="";
outfile="";
outPaircmpfile = "";
pairalnTopoFile = "";
isQuiet=False;
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] == "-h" or sys.argv[i] == "--help":
PrintHelp();
sys.exit();
elif (sys.argv[i] == '-o' or sys.argv[i] == '--o' or sys.argv[i]
== "-outfile" or sys.argv[i] == "--outfile"):
outfile=sys.argv[i+1];
i += 2;
elif sys.argv[i] == "-gap" or sys.argv[i] == "--gap":
parameters['minGapFraction'] = float(sys.argv[i+1]);
i += 2;
elif sys.argv[i] == "-dg" or sys.argv[i] == "--dg":
parameters['maxDGvalue'] = float(sys.argv[i+1]);
i += 2;
elif sys.argv[i] in ["-min-seqidt", "--min-seqidt"]:
parameters['minSeqIDT'] = float(sys.argv[i+1]);
i += 2;
elif sys.argv[i] in ["-max-seqidt", "--max-seqidt"]:
parameters['maxSeqIDT'] = float(sys.argv[i+1]);
i += 2;
elif sys.argv[i] in ["-write-paircmp", "--write-paircmp"]:
outPaircmpfile = sys.argv[i+1];
i += 2;
elif sys.argv[i] in ["-aln", "--aln"]:
pairalnTopoFile = sys.argv[i+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;
if pairalnTopoFile == "":
print >> sys.stderr, "pairalnTopoFile not set. Exit.";
return -1;
pairTopoAlnDict = GetPairTopoAln(pairalnTopoFile);
# pairTopoAlnDict[id1-id2]['id1] ['id2'] ['anno1'] ['anno2] ['seq1']
# ['seq2']
rootname=os.path.basename(os.path.splitext(infile)[0]);
fpout = sys.stdout;
fppaircmp = None;
if outPaircmpfile != "" :
fppaircmp = open(outPaircmpfile, "w")
if outfile != "":
try:
fpout = open(outfile,"w");
except IOError:
print >>sys.stderr, "Failed to write to file %s."%outfile;
print >> sys.stderr, "Reset output to sys.stdout.";
fpout = sys.stdout;
pass;
fpin = open (infile, "rb");
if not fpin:
print >> sys.stderr, "Failed to open input file %s"%(infile);
return -1;
unprocessedBuffer="";
cntTotalOutputRecord = 0;
cntTotalReadInRecord = 0;
isEOFreached = False;
while 1:
buff = fpin.read(BLOCK_SIZE);
if buff == "":
isEOFreached = True;
buff = unprocessedBuffer + buff;
pairCmpRecordList=[];
unprocessedBuffer = lcmp.ReadPairCmpResultFromBuffer(buff,pairCmpRecordList);
if len(pairCmpRecordList) > 0:
#WritePairCmpRecord(pairCmpRecordList,fpout);
filteredList = FilterPairCmpResult(pairCmpRecordList, parameters);
for record in filteredList:
if IsHasInternalVariation(record):
if fppaircmp != None:
li = [];
li.append(record);
(status, cntTotalOutputRecord ) = lcmp.WritePairCmpRecord(li, cntTotalOutputRecord, fppaircmp);
key = "%s-%s"%(record['id1'], record['id2']);
pair = pairTopoAlnDict[key];
fpout.write(">%s\n"%pair['anno1']);
fpout.write("%s\n"%pair['seq1']);
fpout.write(">%s\n"%pair['anno2']);
fpout.write("%s\n"%pair['seq2']);
cntTotalReadInRecord += len(pairCmpRecordList);
if isEOFreached == True:
break;
fpin.close();
print "cntTotalReadInRecord =", cntTotalReadInRecord;
print "cntTotalOutputRecord =", cntTotalOutputRecord;
if fpout != None and fpout != sys.stdout:
fpout.close();
if fppaircmp != None:
fppaircmp.close();
return 0;
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outpath = "./"
dupfile = ""
paircmpfile = ""
outfile = ""
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
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 ["-o", "--o"]:
outfile = argv[i + 1]
i += 2
elif argv[i] in ["-l", "--l"]:
idListFile = argv[i + 1]
i += 2
elif argv[i] in ["-dup", "--dup"]:
dupfile = argv[i + 1]
i += 2
elif argv[i] in ["-paircmp", "--paircmp"]:
paircmpfile = 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:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
if paircmpfile == "":
return 1
if dupfile == "":
return 1
dupPairList = ReadDupPairList(dupfile)
dupPairSet = set(dupPairList)
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
fpin = open(paircmpfile, "r")
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)
if len(pairCmpRecordList) > 0:
StatDupPaircmp(pairCmpRecordList, dupPairSet, fpout)
cntTotalReadInRecord += len(pairCmpRecordList)
if isEOFreached == True:
break
fpin.close()
myfunc.myclose(fpout)
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
i = 1
isNonOptionArg = False
isPickOne = False
paircmpFile = ""
pfamACDEListFile = '/data3/data/pfam/pfamA.seed.ac-delist'
seqDefFile = '/data3/wk/MPTopo/pfamAna/pfam2-selTM-giid-refseqid-pfamid-description.txt'
outpath = ""
htmlname = 'index'
tableinfoFile = ""
while i < numArgv: #{{{
if isNonOptionArg == True:
paircmpFile = argv[i]
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] == "-h" or argv[i] == "--help":
PrintHelp()
return 1
elif argv[i] == "-outpath" or argv[i] == "--outpath":
outpath = argv[i + 1]
i += 2
elif argv[i] == "-htmlname" or argv[i] == "--htmlname":
htmlname = argv[i + 1]
i += 2
elif argv[i] == "-alnfigpath" or argv[i] == "--alnfigpath":
g_params['alnFigPath'] = argv[i + 1]
i += 2
elif argv[i] == "-pfamdef" or argv[i] == "--pfamdef":
pfamACDEListFile = argv[i + 1]
i += 2
elif argv[i] == "-msapath" or argv[i] == "--msapath":
g_params['MSAPath'] = argv[i + 1]
i += 2
elif argv[i] == "-msapath2" or argv[i] == "--msapath2":
g_params['MSAPath2'] = argv[i + 1]
i += 2
elif argv[i] == "-tableinfo" or argv[i] == "--tableinfo":
tableinfoFile = argv[i + 1]
i += 2
elif argv[i] == "-seqdef" or argv[i] == "--seqdef":
seqDefFile = argv[i + 1]
i += 2
elif argv[i] == "-gap" or argv[i] == "--gap":
g_params['minGapFraction'] = float(argv[i + 1])
i += 2
elif argv[i] == "-dg" or argv[i] == "--dg":
g_params['maxDGvalue'] = float(argv[i + 1])
i += 2
elif argv[i] in ["-min-seqidt", "--min-seqidt"]:
g_params['minSeqIDT'] = float(argv[i + 1])
i += 2
elif argv[i] in ["-max-seqidt", "--max-seqidt"]:
g_params['maxSeqIDT'] = float(argv[i + 1])
i += 2
elif argv[i] in ["-tableformat", "--tableformat"]:
g_params['htmltableformat'] = int(argv[i + 1])
i += 2
elif argv[i] in ["-treepath", "--treepath"]:
g_params['treepath'] = argv[i + 1]
i += 2
elif argv[i] in ["-ordermsapath", "--ordermsapath"]:
g_params['ordermsapath'] = argv[i + 1]
i += 2
elif argv[i] in ["-topomsapath", "--topomsapath"]:
g_params['topomsapath'] = argv[i + 1]
i += 2
elif argv[i] in ["-type", "--type"]:
g_params['selecttype'] = argv[i + 1]
i += 2
elif argv[i] in ["-filter-predseq", "--filter-predseq"]:
if argv[i + 1].lower()[0] == 'y':
g_params['isFilterPredictedSeq'] = True
else:
g_params['isFilterPredictedSeq'] = False
i += 2
elif argv[i] == "-q":
isQuiet = True
i += 1
elif argv[i] in ["-pickone", "--pickone"]:
isPickOne = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
paircmpFile = argv[i]
i += 1
#}}}
g_params['outpath'] = outpath
if not os.path.exists(outpath):
os.system("mkdir -p %s" % outpath)
# read paircmprecordlist
fpin = open(paircmpFile, 'r')
buff = fpin.read()
fpin.close()
recordList = []
unprocessedBuffer = lcmp.ReadPairCmpResultFromBuffer(buff, recordList)
print "len(recordList) =", len(recordList)
seqIDListSet = set([])
for record in recordList:
seqIDListSet.add(record['id1'])
seqIDListSet.add(record['id2'])
print "len(seqIDListSet) =", len(seqIDListSet)
# Read In pairwise alignment info
pairalnStat = {}
if tableinfoFile != "" and os.path.exists(tableinfoFile):
pairalnStat = ReadInTableInfo(tableinfoFile)
print "len(pairalnStat) =", len(pairalnStat)
#Read In pfamDefList
if not os.path.exists(pfamACDEListFile):
print >> sys.stderr, "Error! file pfamACDEListFile (%s) does not exist." % pfamACDEListFile
return 1
pfamDefDict = ReadPfamDEList(pfamACDEListFile)
print 'len(pfamDefDict)=', len(pfamDefDict)
#Read in seqinfoList
seqInfoDict = {}
if not os.path.exists(seqDefFile):
print >> sys.stderr, "Error! file seqDefFile (%s) does not exist." % seqDefFile
return 1
fpin = open(seqDefFile, "r")
line = fpin.readline()
line = fpin.readline()
while line:
strs = line.split('|')
if len(strs) == 4:
gid = strs[0].strip()
if gid in seqIDListSet:
refseqid = strs[1].strip()
pfamid = strs[2].strip()
seqdef = strs[3].strip()
seqInfoDict[gid] = {}
seqInfoDict[gid]['pfamid'] = pfamid
seqInfoDict[gid]['refseqid'] = refseqid
seqInfoDict[gid]['seqdef'] = seqdef
seqInfoDict[gid]['pfamdef'] = pfamDefDict[pfamid]
line = fpin.readline()
fpin.close()
print 'len(seqInfoDict)=', len(seqInfoDict)
# add tableinfo to record list
print "Add pairwise alignment table info to record..."
AddTableInfo(recordList, pairalnStat)
print "Add seqdef to record ..."
AddSeqDefInfo(recordList, seqInfoDict)
filteredRecordList = FilterPairCmpResult(recordList)
del recordList
numFilteredRecordList = len(filteredRecordList)
print "numFilteredRecordList = %d" % numFilteredRecordList
# reorder list according to pfamid
numPair = len(filteredRecordList)
tupList = [] # list of (index - pfamid)
for i in xrange(numPair):
thisPfamid = seqInfoDict[filteredRecordList[i]['id1']]['pfamid']
tupList.append((i, thisPfamid))
sorted_by_pfamid = sorted(tupList, key=lambda tup: tup[1])
pairCmpRecordList = []
for i in xrange(numPair):
pairCmpRecordList.append(filteredRecordList[sorted_by_pfamid[i][0]])
if isPickOne:
pairCmpRecordList = PickOnlyOneForEachPfam(pairCmpRecordList,
sorted_by_pfamid)
print "len(pairCmpRecordList) = ", len(pairCmpRecordList)
WriteHTML(pairCmpRecordList, seqInfoDict, htmlname, outpath)
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))