def mkdtst_test(fname, window, answer):
# For test
"""
>>> [i for i in mkdtst_test('.test.pssm',10,answer='./.answer.test')]
None
"""
if answer is not None:
ans_sgr = util.ans(answer)
ans2int = lambda idch, pos: 1 if ans_sgr.isans(idch, pos) else -1
for idch, start, pssm in parse_pssm4pos(fname):
idch = idch.strip()
# 2012/1/31 pos -> pos + start
# !!! now nodyfying !!!
yield idch, {
pos: (ans2int(idch, start + pos), mkvec(pssm, window, pos))
for pos in range(len(pssm))
}
else:
# For negative dataset
for idch, pssm in parse_pssm(fname):
idch = idch.strip()
yield idch, {
pos: (-1, mkvec(pssm, window, pos))
for pos in range(len(pssm))
}
def getpos_fasta(fname,answer):
"""
>>> fname = "../dataset/mono.d4.0.nxg.miss.fasta"
>>> answer = "../dataset/answer_mono.d4.0.nxg.txt"
>>> getpos_fasta(fname,answer).next()
None
"""
# yielding answer position residue
ans_sgr = util.ans(answer)
ans2int = lambda idch,pos: 1 if ans_sgr.isans(idch,pos) else -1
for idch,start,seq in seq2feature.fasta2seq(fname):
idch = idch.strip()
yield idch,{pos + start:res for pos,res in enumerate(seq)
if ans_sgr.isans(idch,start + pos)}
def mkdtst_test(self,window,answer):
# For test
"""
>>> [i for i in mkdtst_test('.test.pssm',10,answer='./.answer.test')]
None
"""
ans_sgr = util.ans(answer)
ans2int = lambda idch,pos: 1 if ans_sgr.isans(idch,pos) else -1
for idch,start,v_pssm in self.parse_pssm4pos():
idch = idch.strip()
idchs = [(idch,pos + start) for pos in range(len(v_pssm))]
dataset = [self.mkvec(v_pssm,window,pos) for pos in range(len(v_pssm))]
label = [ans2int(idch,start + pos) for pos in range(len(v_pssm))]
yield idch,idchs,label,dataset
def mkdtst_test(fname, window, ftr, answer):
# For test
"""
>>> import feature
>>> [i for i in mkdtst_test('.test.fasta',10,lambda seq:feature.seq2frq(seq),answer='../dataset/answer_monod4.0.cluster1.txt')]
None
"""
ans_sgr = util.ans(answer)
ans2int = lambda idch, pos: 1 if ans_sgr.isans(idch, pos) else -1
for idch, start, seq in fasta2seq(fname):
idch = idch.strip()
# 2012/1/31 pos -> pos + start
# !!! now nodyfying !!!
yield idch, {pos: (ans2int(idch, start + pos), mkvec(ftr, seq, window, pos)) for pos in range(len(seq))}
def mkdtst_test4svr(fname, window, answer):
# For test
"""
>>> [i for i in mkdtst_test('.test.pssm',10,answer='./.answer.test')]
None
"""
ans_sgr = util.ans(answer)
for idch, start, pssm in parse_pssm4pos(fname):
idch = idch.strip()
# 2012/1/31 pos -> pos + start
# !!! now nodyfying !!!
yield idch, {
pos: (ans_sgr.get_dist(start + pos,
idch), mkvec(pssm, window, pos))
for pos in range(len(pssm))
}
def mkdtst_train(self,window,answer):
"""
>>> import feature
>>> [i for i in mkdtst_train('.test.pssm',10,answer='.answer.test')]
None
"""
# only binding site For make positive dataet
ans_sgr = util.ans(answer)
# ans2int is function that return +1 if True else return -1
ans2int = lambda idch,pos: 1 if ans_sgr.isans(idch,pos) else -1
for idch,start,v_pssm in self.parse_pssm4pos():
idch = idch.strip()
idchs = [(idch,pos + start) for pos in range(len(v_pssm))]
dataset = [self.mkvec(v_pssm,window,pos) for pos in range(len(v_pssm)) if ans_sgr.isans(idch,start + pos)]
label = [ans2int(idch,start + pos) for pos in range(len(v_pssm)) if ans_sgr.isans(idch,start + pos)]
yield idch,idchs,label,dataset
def mkdtst_neg(self,window,answer,size = 1):
"""
>>> import feature
>>> [i for i in mkdtst_neg('.test.pssm',10,'.answer.test',1)]
None
"""
ans_sgr = util.ans(answer)
# To Dataset randomly choiced positions
for idch,start,pssm in self.parse_pssm4pos():
_indx = [i for i in range(1,len(pssm)) if not ans_sgr.isans(idch,i + start)]
random.shuffle(_indx)
selected_pos = _indx[:size*len(ans_sgr.get_pos(idch))]
idch = idch.strip()
idchs = [(idch,pos + start) for pos in selected_pos]
dataset = [self.mkvec(pssm,window,pos) for pos in selected_pos]
label = [-1 for pos in selected_pos]
yield idch,idchs,label,dataset
def mkdtst_near(fname, window, ftr, answer, low, up):
# For test
"""
>>> [i for i in mkdtst_test('.test.pssm',10,answer='./.answer.test')]
None
"""
ans_sgr = util.ans(answer)
ans2int = lambda idch, pos: 1 if ans_sgr.isans(idch, pos) else -1
for idch, start, seq in fasta2seq(fname):
idch = idch.strip()
# 2012/1/31 pos -> pos + start
# !!! now nodyfying !!!
yield idch, {
pos: (ans2int(idch, start + pos), mkvec(ftr, seq, window, pos))
for pos in range(len(seq))
if low <= ans_sgr.get_dist(pos + start, idch) <= up or ans_sgr.isans(idch, start + pos)
}
def mkdtst_train(fname, window, answer):
"""
>>> import feature
>>> [i for i in mkdtst_train('.test.pssm',10,answer='.answer.test')]
None
"""
# only binding site For make positive dataet
ans_sgr = util.ans(answer)
# ans2int is function that return +1 if True else return -1
ans2int = lambda idch, pos: 1 if ans_sgr.isans(idch, pos) else -1
for idch, start, pssm in parse_pssm4pos(fname):
idch = idch.strip()
# 2012/1/31 pos -> pos + start
# !!! now nodyfying !!!
yield idch, {
pos: (ans2int(idch, start + pos), mkvec(pssm, window, pos))
for pos in range(len(pssm)) if ans_sgr.isans(idch, start + pos)
}
def mkdtst_near(fname, window, answer, low, up):
# For test
"""
>>> [i for i in mkdtst_test('.test.pssm',10,answer='./.answer.test')]
None
"""
ans_sgr = util.ans(answer)
ans2int = lambda idch, pos: 1 if ans_sgr.isans(idch, pos) else -1
for idch, start, pssm in parse_pssm4pos(fname):
idch = idch.strip()
# 2012/1/31 pos -> pos + start
# !!! now nodyfying !!!
yield idch, {
pos: (ans2int(idch, start + pos), mkvec(pssm, window, pos))
for pos in range(len(pssm))
if low <= ans_sgr.get_dist(pos + start, idch) <= up
or ans_sgr.isans(idch, start + pos)
}
def mkdtst_train(fname, window, ftr, answer):
"""
>>> import feature
>>> [i for i in mkdtst_train('.test.fasta',10,lambda seq:feature.seq2frq(seq),answer='../dataset/answer_monod4.0.cluster1.txt')]
None
"""
# only binding site For make positive dataet
ans_sgr = util.ans(answer)
# ans2int is function that return +1 if True else return -1
ans2int = lambda idch, pos: 1 if ans_sgr.isans(idch, pos) else -1
for idch, start, seq in fasta2seq(fname):
idch = idch.strip()
# 2012/1/31 pos -> pos + start
# !!! now nodyfying !!!
yield idch, {
pos: (ans2int(idch, start + pos), mkvec(ftr, seq, window, pos))
for pos in range(len(seq))
if ans_sgr.isans(idch, start + pos)
}
def make_dataset(self,window,answer):
# For test
"""
>>> [i for i in mkdtst_test('.test.pssm',10,answer='./.answer.test')]
None
"""
if answer is not None:
answer = util.ans(answer)
ans2int = lambda idch,pos: 1 if answer.isans(idch,pos) else -1
else:
# if answer is None. this is unknown sample.
ans2int = lambda idch,pos: 0
for idch,v_pssm in self.parse_pssm():
idch = idch.strip()
idchs = [(idch,pos) for pos in range(len(v_pssm))]
dataset = [self.mkvec(v_pssm,window,pos) for pos in range(len(v_pssm))]
label = [ans2int(idch,pos) for pos in range(len(v_pssm))]
yield idch,idchs,label,dataset
def iter_idch2result(idch,name,window,c,g,fans,fname): sgr_ans = util.ans(fans) start = starts(fname)[idch] for pos,val in enumerate(iter_desc(name,window,c,g,idch)): yield pos + start,sgr_ans.isans(idch,pos + start),val