def load_traindata(tfid, args):
print "load_traindata: %s" % tfid
maxrows = 10000 if args.quick else None
trdata = util.load_seq("%s/%s_A.seq.gz" % (datadir, tfid),
minrows=100000,
maxrows=maxrows)
trdata.targetnames = [tfid]
return trdata
def load_traindata(tfid, args):
print "load_traindata: %s"%tfid
maxrows = 10000 if args.quick else None
minrows = 100000
trdata = util.load_seq("%s/%s_AC%s" % (args.indir,tfid,seq_suffix), minrows=minrows, maxrows=maxrows)
trdata.targetnames = [tfid]
if args.mode == "top":
trdata = trdata[:500] # Only top 500 even peaks (A); top 500 odd (B) are stored in the corresponding _B.seq.gz file
elif args.mode == "all":
pass # Top 500 even (A) plus everything else (C)
else:
quit("Unrecognized mode \"%s\". Expected \"top\" or \"all\".")
return trdata
def load_traindata(tfid, args):
print "load_traindata: %s" % tfid
maxrows = 10000 if args.quick else None
minrows = 100000
trdata = util.load_seq("%s/%s_AC%s" % (datadir, tfid, seq_suffix),
minrows=minrows,
maxrows=maxrows)
trdata.targetnames = [tfid]
if args.mode == "top":
trdata = trdata[:
500] # Only top 500 even peaks (A); top 500 odd (B) are stored in the corresponding _B.seq.gz file
elif args.mode == "all":
pass # Top 500 even (A) plus everything else (C)
else:
quit("Unrecognized mode \"%s\". Expected \"top\" or \"all\".")
return trdata
def load_testdata(tedata, tfids, args):
if tedata is not None:
return tedata
if "encode" in datadir:
maxrows = 10000
elif "chip" in datadir:
maxrows = 10000
else:
maxrows = None
all_tedata = {}
for tfid in tfids:
print "load_testdata: %s ..." % tfid,
tedata = util.load_seq("%s/%s_B%s" % (datadir, tfid, seq_suffix), minrows=10000, maxrows=maxrows)
tedata.targetnames = [tfid]
all_tedata[tfid] = tedata
print "done"
return all_tedata
def load_testdata(tedata, tfids, args):
if tedata is not None:
return tedata
if "encode" in datadir:
maxrows = 10000
elif "chip" in datadir:
maxrows = 10000
else:
maxrows = None
all_tedata = {}
for tfid in tfids:
print "load_testdata: %s ..." % tfid,
tedata = util.load_seq("%s/%s_B%s" % (datadir, tfid, seq_suffix),
minrows=10000,
maxrows=maxrows)
tedata.targetnames = [tfid]
all_tedata[tfid] = tedata
print "done"
return all_tedata
def load_traindata(tfid, args):
print "load_traindata: %s"%tfid
maxrows = 10000 if args.quick else None
trdata = util.load_seq("%s/%s_A.seq.gz" % (datadir,tfid), minrows=100000, maxrows=maxrows)
trdata.targetnames = [tfid]
return trdata