def main(g_params):#{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
outfile = ""
infile = ""
mapfile = ""
i = 1
isNonOptionArg=False
while i < numArgv:
if isNonOptionArg == True:
infile = argv[i]
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-o", "--o", "-outfile"]:
(outfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-map", "--map", "-mapfile"]:
(mapfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
infile = argv[i]
i += 1
if myfunc.checkfile(infile) != 0:
return 1
if myfunc.checkfile(mapfile) != 0:
return 1
clanid2pfamidDict = myfunc.ReadFam2SeqidMap(mapfile)
pfamPercentTMDict = ReadPercentTM(infile)
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
GetPercentTMOfClan(pfamPercentTMDict, clanid2pfamidDict, fpout)
myfunc.myclose(fpout)
def main(g_params): #{{{
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
try:
subprocess.check_output(["mkdir", "-p", outpath])
except subprocess.CalledProcessError, e:
print e
return 1
if myfunc.checkfile(pfamid2seqidFile, "pfamid2seqidFile") != 0:
return 1
if myfunc.checkfile("%s0.db" % topodb, "topodb") != 0:
return 1
if myfunc.checkfile("%s0.db" % seqdb, "seqdb") != 0:
return 1
pfamid2seqidDict = myfunc.ReadFam2SeqidMap(pfamid2seqidFile)
hdl_topo = myfunc.MyDB(topodb)
if not hdl_topo.failure:
idSet_topo = set(hdl_topo.indexedIDList)
else:
idSet_topo = set([])
print >> sys.stderr, "Failed to open topology database %s" % (topodb)
return 1
hdl_seq = myfunc.MyDB(seqdb)
if hdl_seq.failure:
print >> sys.stderr, "Failed to open sequence database %s" % (seqdb)
return 1
GetTMProList_per_family(pfamid2seqidDict, idSet_topo, hdl_seq, hdl_topo,
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
infile = ""
outfile = ""
seqid2pfamidFile = datadir3 + os.sep + "wk/MPTopo/pfamAna_refpro/pfammap_from_uniprot/refpro20120604-celluar.selmaxlength-m1.nr100.seqid2pfamid"
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 ["-seqid2pfamid", "--seqid2pfamid"]:
seqid2pfamidFile = argv[i + 1]
i += 2
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("Error! Wrong argument:", argv[i], file=sys.stderr)
return 1
else:
infile = argv[i]
i += 1
if infile == "" or not os.path.exists(infile):
print("Error. Infile not set. exit", file=sys.stderr)
return 1
if seqid2pfamidFile == "" or not os.path.exists(seqid2pfamidFile):
print("Error. seqid2pfamidFile does not exist. exit", file=sys.stderr)
return 1
seqid2pfamidDict = myfunc.ReadFam2SeqidMap(seqid2pfamidFile)
if seqid2pfamidFile == {}:
print("Read seqid2pfamidFile failed.", file=sys.stderr)
return 1
(idList, topoList) = myfunc.ReadFasta_without_annotation(infile)
if len(idList) < 1:
print("Read infile failed.", file=sys.stderr)
return 1
idList.remove("Consensus")
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
WritePfamColorDef(idList, seqid2pfamidDict, fpout)
myfunc.myclose(fpout)
return 0
def main(g_params): #{{{
argv = sys.argv
numArgv = len(argv)
if numArgv < 2:
PrintHelp()
return 1
SPE_PAIR_LIST = [(2, 1), (2, 4), (2, 6), (2, 8), (3, 6), (3, 7), (4, 6),
(4, 8), (4, 10), (5, 7), (5, 10), (6, 8), (6, 10),
(6, 12), (7, 14), (8, 10), (8, 12), (10, 12), (10, 13),
(11, 13), (12, 14)]
outfile = ""
infile = ""
pfamDefFile = "%s/data/pfam/pfam26.0/Pfam-A.clans.tsv" % (DATADIR3)
signalpFile = "%s/wk/MPTopo/pfamAna_refpro/pred_signalp/refpro20120604-celluar.selmaxlength-m1.nr100.signalp_list" % (
DATADIR3)
#seqid2clanidMapFile = "%s/wk/MPTopo/pfamAna_refpro/pfammap_from_uniprot/refpro20120604-celluar.selmaxlength-m1.nr100.filter.fragmented.seqid2clanid"%(DATADIR3)
#seqid2pfamidMapFile = "%s/wk/MPTopo/pfamAna_refpro/pfammap_from_uniprot/refpro20120604-celluar.selmaxlength-m1.nr100.filter.fragmented.seqid2pfamid"%(DATADIR3)
seqid2clanidMapFile = ""
seqid2pfamidMapFile = ""
tm_pfamidListFile = ""
tm_clanidListFile = ""
pfamid2seqidMapFile = ""
clanid2seqidMapFile = ""
dbname_predTM = ""
pairlistwithpfamidFile = ""
pfamtype = ""
pairListFile = ""
#classList_TableNumTMHeatMap = ["ALL", "RMSP"]
classList_TableNumTMHeatMap = ["ALL"]
i = 1
isNonOptionArg = False
while i < numArgv:
if isNonOptionArg == True:
infile = argv[i]
isNonOptionArg = False
i += 1
elif argv[i] == "--":
isNonOptionArg = True
i += 1
elif argv[i][0] == "-":
if argv[i] in ["-h", "--help"]:
PrintHelp()
return 1
elif argv[i] in ["-o", "--o", "-outfile"]:
(outfile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-outpath", "--outpath"]:
(g_params['outpath'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-l", "--l"]:
(fileListFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-pfamdef", "--pfamdef"]:
(pfamDefFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-signalp", "--signalp"]:
(signalpFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-mp", "--mp"]:
g_params[
'pairwise_comparison_method'], i = myfunc.my_getopt_int(
argv, i)
elif argv[i] in ["-mindiffpair", "--mindiffpair"]:
g_params['mindiffpair'], i = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-pfamtype", "--pfamtype"]:
pfamtype, i = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-clanidlist", "--clanidlist"]:
(tm_clanidListFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-pfamidlist", "--pfamidlist"]:
(tm_pfamidListFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-seqid2clanid", "--seqid2clanid"]:
(seqid2clanidMapFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-seqid2pfamid", "--seqid2pfamid"]:
(seqid2pfamidMapFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-pfamid2seqid", "--pfamid2seqid"]:
(pfamid2seqidMapFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-clanid2seqid", "--clanid2seqid"]:
(clanid2seqidMapFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-pairlistwithpfamid", "--pairlistwithpfamid"]:
(pairlistwithpfamidFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-predTMdbname", "--predTMdbname"]:
(dbname_predTM, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-pairlist", "--pairlist"]:
(pairListFile, i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-winsize", "--winsize"]:
(g_params['winsize'], i) = myfunc.my_getopt_int(argv, i)
elif argv[i] in ["-outname", "--outname"]:
(g_params['outname'], i) = myfunc.my_getopt_str(argv, i)
elif argv[i] in ["-q", "--q"]:
g_params['isQuiet'] = True
i += 1
elif argv[i] in ["-prokar", "--prokar"]:
g_params['isOnlyAnaProkar'] = True
i += 1
elif argv[i] in ["-eukar", "--eukar"]:
g_params['isOnlyAnaEukar'] = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
infile = argv[i]
i += 1
if myfunc.checkfile(
infile, "%s (line %d): infile" %
(__file__, inspect.currentframe().f_lineno)) != 0:
return 1
dirpath = myfunc.my_dirname(infile)
# try to obtain Pfam family tag
tag = ""
if pfamtype != "":
if pfamtype.upper().find("FAM") != -1:
tag = ".Family"
elif pfamtype.upper().find("DOM") != -1:
tag = ".Domain"
elif pfamtype.upper().find("REP") != -1:
tag = ".Repeat"
elif pfamtype.upper().find("MOT") != -1:
tag = ".Motif"
else:
tag = ""
else:
if infile.find(".Family.") != -1:
tag = ".Family"
elif infile.find(".Domain.") != -1:
tag = ".Domain"
elif infile.find(".Repeat.") != -1:
tag = ".Repeat"
elif infile.find(".Motif.") != -1:
tag = ".Motif"
else:
tag = ""
if seqid2clanidMapFile == "":
seqid2clanidMapFile = "%s/wk/MPTopo/pfamAna_refpro/pfammap_from_uniprot/Pfam-A-full.seqfrompfamfasta.percentTMpro_scampi.perTM75_nseq20.nr100.filter.fragmented.seqid2clanid" % (
DATADIR3)
if myfunc.checkfile(
seqid2clanidMapFile, "%s (line %d): seqid2clanidMapFile" %
(__file__, inspect.currentframe().f_lineno)):
return 1
if seqid2pfamidMapFile == "":
seqid2pfamidMapFile = "%s/wk/MPTopo/pfamAna_refpro/pfammap_from_uniprot/Pfam-A-full.seqfrompfamfasta.percentTMpro_scampi.perTM75_nseq20%s.nr100.filter.fragmented.seqid2pfamid" % (
DATADIR3, tag)
if myfunc.checkfile(
seqid2pfamidMapFile, "%s (line %d): seqid2pfamidMapFile" %
(__file__, inspect.currentframe().f_lineno)):
return 1
if pfamid2seqidMapFile == "":
pfamid2seqidMapFile = "%s/wk/MPTopo/pfamAna_refpro/pfammap_from_uniprot/Pfam-A-full.seqfrompfamfasta.percentTMpro_scampi.perTM75_nseq20.nr100.filter.fragmented.pfamid2seqid" % (
DATADIR3)
if myfunc.checkfile(
pfamid2seqidMapFile, "%s (line %d): pfamid2seqidMapFile" %
(__file__, inspect.currentframe().f_lineno)):
return 1
if clanid2seqidMapFile == "":
clanid2seqidMapFile = "%s/wk/MPTopo/pfamAna_refpro/pfammap_from_uniprot/Pfam-A-full.seqfrompfamfasta.percentTMpro_scampi.perTM75_nseq20%s.nr100.filter.fragmented.clanid2seqid" % (
DATADIR3, tag)
if myfunc.checkfile(
clanid2seqidMapFile, "%s (line %d): clanid2seqidMapFile" %
(__file__, inspect.currentframe().f_lineno)):
return 1
if tm_pfamidListFile == "":
tm_pfamidListFile = "%s/data/pfam/pfam26.0/Pfam-A-full.seqfrompfamfasta.percentTMpro_scampi.perTM75_nseq20%s.pfamidlist" % (
DATADIR3, tag)
if myfunc.checkfile(
tm_pfamidListFile, "%s (line %d): tm_pfamidListFile" %
(__file__, inspect.currentframe().f_lineno)):
return 1
if tm_clanidListFile == "":
tm_clanidListFile = "%s/data/pfam/pfam26.0/Pfam-A-full.seqfrompfamfasta.percentTMpro_scampi.perTM75_nseq20.clanidlist" % (
DATADIR3)
if myfunc.checkfile(
tm_clanidListFile, "%s (line %d): tm_clanidListFile" %
(__file__, inspect.currentframe().f_lineno)):
return 1
if dbname_predTM == "":
dbname_predTM = "%s/wk/MPTopo/pfamAna_refpro/pred_topcons_single_method4/refpro20120604-celluar.selmaxlength-m1.topcons-single_topcons_single.m1.agree-44.RMSP" % (
DATADIR3)
if myfunc.checkfile(
"%s0.db" % (dbname_predTM), "%s (line %d): dbname_predTM" %
(__file__, inspect.currentframe().f_lineno)):
return 1
if g_params['isOnlyAnaProkar']:
prokarseqidfile = "%s/data/uniprot/reference_proteome/refpro20120604-celluar.selmaxlength-m1.nr100.filter.fragmented.Prokaryota.seqidlist" % (
DATADIR3)
g_params['prokarSeqIDSet'] = set(myfunc.ReadIDList(prokarseqidfile))
if len(g_params['prokarSeqIDSet']) < 1:
return 1
if g_params['isOnlyAnaEukar']:
eukarseqidfile = "%s/data/uniprot/reference_proteome/refpro20120604-celluar.selmaxlength-m1.nr100.filter.fragmented.Eukaryota.seqidlist" % (
DATADIR3)
g_params['eukarSeqIDSet'] = set(myfunc.ReadIDList(eukarseqidfile))
if len(g_params['eukarSeqIDSet']) < 1:
return 1
if pairlistwithpfamidFile == "":
pairlistwithpfamidFile = "%s/../../Pfam-.maxpair100.pairlistwithpfamid" % (
dirpath)
if myfunc.checkfile(
pairlistwithpfamidFile, "%s (line %d): pairlistwithpfamidFile" %
(__file__, inspect.currentframe().f_lineno)):
return 1
pfamid_2_seqidpair_Dict = ReadPairListWithFamID(pairlistwithpfamidFile)
usedPfamIDSet = set(
pfamid_2_seqidpair_Dict.keys()) # pfamids used in pair selection
if pairListFile != "":
li = myfunc.ReadPairList(pairListFile)
SPE_PAIR_LIST = []
for tup in li:
SPE_PAIR_LIST.append((int(tup[0]), int(tup[1])))
(pfamidDefDict, clanidDefDict) = ReadPfamDefFile(pfamDefFile)
signalpDict = lcmp.ReadSignalPDict(signalpFile)
seqid2clanidDict = myfunc.ReadFam2SeqidMap(seqid2clanidMapFile)
seqid2pfamidDict = myfunc.ReadFam2SeqidMap(seqid2pfamidMapFile)
clanid2seqidDict = myfunc.ReadFam2SeqidMap(clanid2seqidMapFile)
pfamid2seqidDict = myfunc.ReadFam2SeqidMap(pfamid2seqidMapFile)
tm_pfamidList = myfunc.ReadIDList(tm_pfamidListFile)
tm_clanidList = myfunc.ReadIDList(tm_clanidListFile)
tm_pfamidSet = set(tm_pfamidList)
tm_clanidSet = set(tm_clanidList)
hdl_predTM = myfunc.MyDB(dbname_predTM)
if not hdl_predTM.failure:
idSet_TMpro = set(hdl_predTM.indexedIDList)
else:
idSet_TMpro = set([])
#classList_TableNumTMHeatMap = ["ALL", "RMSP", "RMDUP"]
#alignrangeList = ['FULL_ALIGNED', 'all', 'PART_ALIGNED']
alignrangeList = ['FULL_ALIGNED']
if g_params['outpath'] != "" and not os.path.exists(g_params['outpath']):
cmd = ["mkdir", "-p", g_params['outpath']]
try:
subprocess.check_call(cmd)
except subprocess.CalledProcessError, e:
print e
return 1
def main(g_params):#{{{
argv = sys.argv
numArgv=len(argv)
if numArgv < 2:
PrintHelp()
return 1
i = 1
isNonOptionArg=False
isPickOne = False
infile = ""
pfamDefFile = '/data3/data/pfam/pfam26.0/Pfam-A.clans.tsv'
seqDefFile = '/data3/wk/MPTopo/pfamAna/pfam2-selTM-giid-refseqid-pfamid-description.txt'
idwithannoFile = "/data3/wk/MPTopo/pfamAna_refpro/pfammap_from_uniprot/Pfam-A-full.perTM75_nseq20.nr100.filter.fragmented.uniq.idwithanno"
seqLengthFile = "/data3/wk/MPTopo/pfamAna_refpro/pfammap_from_uniprot/Pfam-A-full.perTM75_nseq20.nr100.filter.fragmented.uniq.seqlen"
topoDB = "/data3/wk/MPTopo/pfamAna_refpro/pred_topcons/refpro20120604-celluar.selmaxlength-m1.topcons.result_TOPCONS.topo"
seqDB = "/data3/wk/MPTopo/pfamAna_refpro/cellular_filter_fragment/Pfam-A-full.perTM75_nseq20.nr100.filter.fragmented.uniq"
pfamscanFile = "/data3/wk/MPTopo/pfamAna_refpro/result_pfamscan/Pfam-A-full.perTM75_nseq20.nr100.filter.fragmented.pfamscan"
outpath = ""
outfile = ""
htmlname = 'index'
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] == "-h" or argv[i] == "--help":
PrintHelp()
return 1
elif argv[i] in ["-outpath", "--outpath"]:
outpath = argv[i+1]
i += 2
elif argv[i] in ["-o", "--o"]:
(outfile, i) = myfunc.my_getopt_str(argv,i)
elif argv[i] in ["-htmlname", "--htmlname"]:
htmlname = argv[i+1]
i += 2
elif argv[i] in ["-seqlen", "--seqlen"]:
seqLengthFile = argv[i+1]
i += 2
elif argv[i] in ["-topodb", "--topodb"]:
topoDB = argv[i+1]
i += 2
elif argv[i] in ["-seqdb", "--seqdb"]:
seqDB = argv[i+1]
i += 2
elif argv[i] in [ "-pfamdef", "--pfamdef"]:
pfamDefFile = argv[i+1]
i += 2
elif argv[i] in ["-pfamscan", "--pfamscan"]:
pfamscanFile = argv[i+1]
i += 2
elif argv[i] in ["-seqdef", "--seqdef"]:
seqDefFile = argv[i+1]
i += 2
elif argv[i] in ["-q"]:
isQuiet = True
i += 1
else:
print >> sys.stderr, "Error! Wrong argument:", argv[i]
return 1
else:
infile = argv[i]
i += 1
#}}}
g_params['outpath'] = outpath
if outpath == "":
print >> sys.stderr, "outpath not set"
return 1
elif outpath != "" and not os.path.exists(outpath):
os.makedirs(outpath)
if infile == "":
print >> sys.stderr, "infile not set"
return 1
fpout = myfunc.myopen(outfile, sys.stdout, "w", False)
seqid2pfamidDict = myfunc.ReadFam2SeqidMap(infile)
# group sequences by domains
groupDict = {}
for seqid in seqid2pfamidDict:
famlist = seqid2pfamidDict[seqid]
ss = "\t".join(famlist)
if not ss in groupDict:
groupDict[ss] = []
groupDict[ss].append(seqid)
groupList = []
for ss in groupDict:
groupList.append((ss, len(groupDict[ss]), groupDict[ss]))
groupList = sorted(groupList, key=lambda x:x[1], reverse=True)
seqlenDict = ReadSeqLengthDict(seqLengthFile)
seqannoDict = ReadIDWithAnnoInfo(idwithannoFile)
(pfamidDefDict, clanidDefDict) = lcmp.ReadPfamDefFile(pfamDefFile)
topoDict = GetTopoDict(topoDB, seqid2pfamidDict.keys())
seqDict = GetTopoDict(seqDB, seqid2pfamidDict.keys())
pfamScanDict = myfunc.ReadPfamScan(pfamscanFile)
groupedPfamScanDict = GroupPfamScanDict(pfamScanDict)
# WriteHTML(seqid2pfamidDict, seqlenDict, seqannoDict, pfamidDefDict,
# clanidDefDict, topoDict, htmlname, outpath)
WriteInfo(groupList, seqlenDict, seqannoDict, pfamidDefDict,
clanidDefDict, topoDict, groupedPfamScanDict, htmlname, fpout)
MakeAlignment(groupList, seqDict, topoDict, outpath)
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))