def extract_windows(sequences, bg, location=None, wl=None, params=None):
location = location or bg.location
r = count_words(sequences,
bg.l,
location,
strand=bg.strand,
overlap=bg.overlap,
params=params)
R = []
H = r['H']
N = r['N']
#print H
wl = wl or list(H.keys())
info = cli.Info(len(wl), 1, 1)
for w in wl:
info('Extracting windows')
# if params['heuristic'] == 'score':
# x = extract_windows_score(N, H, bg, location=location, w=w, params=params)
# else:
extractor = Extractor(N, H, bg, location=location, w=w, params=params)
x = extractor.run()
R.extend(x)
#print pbinom_cached.stats()
return {'R': R, 'scannedWords': r['scannedWords']}
def count_words_hash(sequences, l, searchLocation, strand='+-', overlap=False):
"""Count each word of length l in sequences
l -- oligonucleotide length
searchLocation -- location tuple example (-200,-1)
strand -- + or +-
overlap -- allow auto-overlapping
return N, H
"""
location = find_location(sequences)
H = {} #hash table key=oligonucleotide value=list of occurrence position
N = {} #scanned bases per position for each word size
N[l] = [0] * (searchLocation[1] - searchLocation[0] + 1)
scannedPositions = 0
scannedWords = 0
info = cli.Info(len(sequences), 1, 1)
for s in sequences:
info('Counting words in [%+05d:%+05d]' %
(searchLocation[0], searchLocation[1]))
dna = s.sequence
HS = {} # for current sequence
a, b = max(searchLocation[0],
s.location[0]), min(searchLocation[1],
s.location[1] - l + 1)
for I in range(a, b + 1):
scannedPositions += 1
i = I - s.location[0]
if dna[i:i + l].find('N') >= 0:
continue
if strand == '+':
w = dna[i:i + l]
elif strand == '+-':
wf = dna[i:i + l]
wr = reverse_complement(dna[i:i + l])
w = min(wf, wr)
N[l][I - searchLocation[0]] += 1
if not overlap and HS.get(w, [a - l])[-1] + l > I:
continue
H.setdefault(w, []).append(I)
HS.setdefault(w, []).append(I)
scannedWords += 1
return dict(N=N,
H=H,
scannedPositions=scannedPositions,
scannedWords=scannedWords)
def realtest(N=2):
#filename = 'E2F.fa'
filename = 'Test/MM0.fa'
sequences = dna.fasta2sequences(filename)
print(sequences)
st = SuffixTree(maxDepth=N,
overlapping=True,
maxIUPAC=N,
NExtension=(1, 1),
storePosition=0)
info = cli.Info(len(sequences))
for i in range(len(sequences)):
info('Processing sequence %s' % sequences[i].id)
st.add_dna(sequences[i].sequence)
#display(st.root)
#count = group_rc(st.count(minLength=N, maxLength=N))
#print_count(count)
return st
def build_markov(self, sequences, order=1):
info = cli.Info((self.location[1] - self.l + 1 - self.location[0]) /
self.step + 1)
for i in range(self.location[0], self.location[1] - self.l + 1,
self.step):
info('building Markov BG pos=%i' % i)
loc = (i, i + self.W - 1)
mm = MM(order)
mm.dyad = self.params['dyad']
mm.monad = self.l
mm.overlap = self.overlap
mm.strand = self.strand
mm.set_NExtension(self.params['spacing'])
mm.learn([s.get_dna(loc) for s in sequences])
self[loc] = mm
sorted_keys = list(self.keys())
sorted_keys.sort()
self.intervals = sorted_keys
def build(self, sequences):
info = cli.Info((self.location[1] - self.l + 1 - self.location[0]) /
self.step + 1)
for i in range(self.location[0], self.location[1] - self.l + 1,
self.step):
loc = (i, i + self.W - 1)
info('building BG pos=%i' % i)
r = count_words(sequences,
self.l,
loc,
strand=self.strand,
overlap=self.overlap,
params=self.params)
H = r['H']
N = r['N']
#compute frequency
for k in H:
H[k] = len(H[k]) / float(sum(N[len(k)]))
self[loc] = H
sorted_keys = list(self.keys())
sorted_keys.sort()
self.intervals = sorted_keys
def add_sequences(self, sequences):
info = cli.Info(len(sequences))
for i in range(len(sequences)):
info('Proccessing sequence %s' % sequences[i].id)
self.add_dna(sequences[i].sequence)
def count_words_tree(sequences,
l,
searchLocation,
strand='+-',
overlap=False,
error=0,
spacing=(1, 1)):
"""Count each word of length l in sequences
l -- oligonucleotide length
searchLocation -- location tuple example (-200,-1)
strand -- + or +-
overlap -- allow auto-overlapping
return N, H
"""
location = find_location(sequences)
#H = {} #hash table key=oligonucleotide value=list of occurrence position
N = {} #scanned base count per position fo each word size
N[l] = [0] * (searchLocation[1] - searchLocation[0] + 1)
scannedPositions = 0
scannedWords = 0
info = cli.Info(len(sequences), 1, 1)
#
# construct SuffixTree
#
st = ST.SuffixTree(maxDepth=l,
overlapping=overlap,
maxIUPAC=error,
NExtension=spacing,
storePosition=True)
for s in sequences:
info('Counting words in [%+05d:%+05d]' %
(searchLocation[0], searchLocation[1]))
a, b = max(searchLocation[0],
s.location[0]), min(searchLocation[1],
s.location[1] - l + 1)
dna = s.get_dna((a, b + l + 1))
st.add_dna(dna, shift=a)
for I in range(a, b + 1):
i = I - s.location[0]
w = dna[i:i + l]
if w.find('N') >= 0:
continue
N[l][I - searchLocation[0]] += 1
#@DEBUG
#ST.display(st.root, maxDepth=6, full=1)
#
# Count
#
#@DEBUG
#keys = st.extract(minLength=l, maxLength=l).keys()
#keys.sort()
#print '\n'.join(keys)
C = st.extract(minLength=l, maxLength=l)
if strand == '+-':
H = ST.get_positions_two_strands(C, overlap)
else:
H = ST.get_positions(C)
return dict(N=N,
H=H,
scannedPositions=scannedPositions,
scannedWords=scannedWords)
def count_dyads_hash(sequences,
l,
spacing,
searchLocation,
strand='+-',
overlap=False):
"""Count each dyad of length l in sequences
l -- oligonucleotide length
spacing -- spacing
searchLocation -- location tuple example (-200,-1)
strand -- + or +-
overlap -- allow auto-overlapping
"""
lmonad = l
l = 2 * lmonad + spacing
location = find_location(sequences)
H = {}
N = {}
N[l] = [0] * (searchLocation[1] - searchLocation[0] + 1)
scannedPositions = 0
scannedWords = 0
info = cli.Info(len(sequences), 1, 1)
for s in sequences:
info('Counting words in [%+05d:%+05d]' %
(searchLocation[0], searchLocation[1]))
dna = s.sequence
HS = {}
a, b = max(searchLocation[0],
s.location[0]), min(searchLocation[1],
s.location[1] - l + 1)
for I in range(a, b + 1):
scannedPositions += 1
i = I - s.location[0]
if dna[i:i + l].find('N') >= 0:
continue
if strand == '+':
w = dna[i:i +
lmonad] + 'N' * spacing + dna[i + lmonad + spacing:i +
2 * lmonad + spacing]
elif strand == '+-':
wf = dna[i:i +
lmonad] + 'N' * spacing + dna[i + lmonad + spacing:i +
2 * lmonad + spacing]
wr = reverse_complement(wf)
w = min(wf, wr)
N[l][I - searchLocation[0]] += 1
if not overlap and HS.get(w, [a - l])[-1] + l > I:
continue
H.setdefault(w, []).append(I)
HS.setdefault(w, []).append(I)
scannedWords += 1
return dict(N=N,
H=H,
scannedPositions=scannedPositions,
scannedWords=scannedWords)