def _mkneg(fname,window,ftr,pssm = False):
with open(fname) as fp:
neg = [i[0] for i in mkftr.getseq(fp)]
# ? didn't idchoice() need ?
for cnt,group in enumerate(mkftr.idchoice(neg)):
fp.seek(0)
for negid,seq in mkftr.getseq(fp):
if pssm:
# For PSSM
seq = [int(i) for i in seq.split()]
if len(seq) < 100:
continue
else:
if len(seq) < 100 or seq.find('X') > 0:
continue
if negid in group:
_indx = [i for i in range(1,len(seq))]
random.shuffle(_indx)
yield negid,{pos:(-1,mkvec(ftr,seq,window,pos,pssm = pssm)) for pos in sorted(_indx[:5])}
def _mkdtst(window,ftr,represent = 'dataset/sugar.txt',
answer = 'dataset/answer_sugar.txt',
name = 'sugar',
flg = True,pssm = False):
def ans2int(ans,idch,pos):
if ans.isans(idch,pos):
return 1
return -1
with open(represent) as fp:
lid = [line.strip() for line in iter(fp.readline,'')]
ans_sgr = mkftr.ans(answer)
# ? Dose idchoice need ?
for cnt,group in enumerate(mkftr.idchoice(lid)):
for idch,start,seq in mkftr.slider(group,name,pssm = pssm):
idch = idch.strip()
# 2012/1/31 pos -> pos + start
# !!! now nodyfying !!!
yield idch,{pos:(ans2int(ans_sgr,idch,start + pos),mkvec(ftr,seq,window,pos,pssm = pssm))
for pos in range(len(seq)) if ans_sgr.isans(idch,start + pos) or not flg}
# !!! now modfying !!!
"""
