def __init__(self, *args):
unittest.TestCase.__init__(self, *args)
fname = "data/test_data/sequences/alignment.fa"
dl = DataLoader()
_, self.seq_x, self.ann_x, self.seq_y, self.ann_y = dl.loadSequence(fname)
self.seq_xs = Fasta.alnToSeq(self.seq_x)
self.seq_ys = Fasta.alnToSeq(self.seq_y)
def _classification(self, sequence_x, ann_x, sequence_y, ann_y):
def state(i):
if sequence_x[i] == '-' and sequence_y[i] == '-':
return 0
if sequence_x[i] == '-':
return 2
if sequence_y[i] == '-':
return 1
return 0
def get_pos():
def state(i):
if sequence_x[i] == '-' and sequence_y[i] == '-':
return -1
if sequence_x[i] == '-':
return 2
if sequence_y[i] == '-':
return 1
return 0
pos_x, pos_y = 0, 0
pos = list()
for i in xrange(len(sequence_x)):
pos.append((pos_x, pos_y))
s = state(i)
if s == 0:
pos_x += 1
pos_y += 1
if s == 1:
pos_x += 1
if s == 2:
pos_y += 1
return pos
assert len(sequence_y) == len(sequence_x)
l = len(sequence_x)
sequence_xs = Fasta.alnToSeq(sequence_x)
sequence_ys = Fasta.alnToSeq(sequence_y)
positions = get_pos()
ret_match = (
0 for _ in filter(lambda x: state(x) == 0, xrange(l))
)
ret_insertX = self.clf.multi_prepare_predict(
(sequence_xs, positions[pos][0], ann_x, sequence_ys, positions[pos][1], ann_y)
for pos in filter(lambda x: state(x) == 1, xrange(l))
)
ret_insertY = self.clf.multi_prepare_predict(
(sequence_ys, positions[pos][1], ann_y, sequence_xs, positions[pos][0], ann_x)
for pos in filter(lambda x: state(x) == 2, xrange(l))
)
ret = merge((state(x) for x in xrange(l)), ret_match, ret_insertX, ret_insertY)
return ret
def __init__(self, *args):
unittest.TestCase.__init__(self, *args)
fname = "data/test_data/sequences/alignment.fa"
dl = DataLoader()
_, self.seq_x, self.ann_x, self.seq_y, self.ann_y = dl.loadSequence(
fname)
self.seq_xs = Fasta.alnToSeq(self.seq_x)
self.seq_ys = Fasta.alnToSeq(self.seq_y)
def _classification(self, sequence_x, ann_x, sequence_y, ann_y):
def state(i):
if sequence_x[i] == '-' and sequence_y[i] == '-':
return 0
if sequence_x[i] == '-':
return 2
if sequence_y[i] == '-':
return 1
return 0
def get_pos():
def state(i):
if sequence_x[i] == '-' and sequence_y[i] == '-':
return -1
if sequence_x[i] == '-':
return 2
if sequence_y[i] == '-':
return 1
return 0
pos_x, pos_y = 0, 0
pos = list()
for i in xrange(len(sequence_x)):
pos.append((pos_x, pos_y))
s = state(i)
if s == 0:
pos_x += 1
pos_y += 1
if s == 1:
pos_x += 1
if s == 2:
pos_y += 1
return pos
assert len(sequence_y) == len(sequence_x)
l = len(sequence_x)
sequence_xs = Fasta.alnToSeq(sequence_x)
sequence_ys = Fasta.alnToSeq(sequence_y)
positions = get_pos()
ret_match = (0 for _ in filter(lambda x: state(x) == 0, xrange(l)))
ret_insertX = self.clf.multi_prepare_predict(
(sequence_xs, positions[pos][0], ann_x, sequence_ys,
positions[pos][1], ann_y)
for pos in filter(lambda x: state(x) == 1, xrange(l)))
ret_insertY = self.clf.multi_prepare_predict(
(sequence_ys, positions[pos][1], ann_y, sequence_xs,
positions[pos][0], ann_x)
for pos in filter(lambda x: state(x) == 2, xrange(l)))
ret = merge((state(x) for x in xrange(l)), ret_match, ret_insertX,
ret_insertY)
return ret
def main(input_files, output_file):
global width
alignments = [
list(
Fasta.load(name, '', Alignment,
['^sequence1', '^sequence2', '^[av].*']))
if os.path.exists(name) else None for name in input_files
]
x_len = len(Fasta.alnToSeq(alignments[0][0].sequences[0]))
y_len = len(Fasta.alnToSeq(alignments[0][0].sequences[1]))
I = Image.new('RGB', (x_len * width + 50, y_len * width + 50),
(255, 255, 255))
D = ImageDraw.Draw(I)
colors = [(0, 0, 0), (255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 0, 255),
(0, 255, 255)]
i = -1
for aln in alignments:
i += 1
if aln == None:
continue
aln = list(aln)
if len(aln) == 0:
continue
aln = aln[0]
try:
annotation = aln.sequences[2]
coords = aln.getCoordPairs()
print coords
x_shift = width / 2 + 25 + i
y_shift = width / 2 + 25 + i * 2
D.line([(x * width + x_shift, y * width + y_shift)
for x, y, _ in coords],
fill=colors[i])
if annotation != None:
for x, y, ind in coords:
if annotation[ind] != 'R':
continue
D.rectangle([(x * width + x_shift - width / 4,
y * width + y_shift - width / 4),
(x * width + x_shift + width / 4,
y * width + y_shift + width / 4)],
outline=colors[i])
except IndexError:
pass
except IOError:
pass
del D
I.save(output_file)
def prepare_training_data(
self,
sequence_x,
annotations_x,
sequence_y,
annotations_y,
):
"""Takes sequences with spaces and prepares training data for classifier
"""
assert len(sequence_y) == len(sequence_x)
sequence_xs = Fasta.alnToSeq(sequence_x)
sequence_ys = Fasta.alnToSeq(sequence_y)
train_data1, matched_pos, seq_size, weights_set = self.prepare_positive_data(
sequence_x,
sequence_xs,
annotations_x,
sequence_y,
sequence_ys,
annotations_y,
)
train_data0 = self.prepare_negative_data(
sequence_x,
sequence_xs,
annotations_x,
sequence_y,
sequence_ys,
annotations_y,
matched_pos,
seq_size,
weights_set,
)
return train_data1[0] + train_data0[0],\
train_data1[1] + train_data0[1],\
train_data1[2] + train_data0[2]
def prepare_training_data(
self,
sequence_x,
annotations_x,
sequence_y,
annotations_y,
):
"""Takes sequences with spaces and prepares training data for classifier
"""
assert len(sequence_y) == len(sequence_x)
sequence_xs = Fasta.alnToSeq(sequence_x)
sequence_ys = Fasta.alnToSeq(sequence_y)
train_data1, matched_pos, seq_size, weights_set = self.prepare_positive_data(
sequence_x,
sequence_xs,
annotations_x,
sequence_y,
sequence_ys,
annotations_y,
)
train_data0 = self.prepare_negative_data(
sequence_x,
sequence_xs,
annotations_x,
sequence_y,
sequence_ys,
annotations_y,
matched_pos,
seq_size,
weights_set,
)
return train_data1[0] + train_data0[0],\
train_data1[1] + train_data0[1],\
train_data1[2] + train_data0[2]
def main(input_files, output_file):
global width
alignments = [list(Fasta.load(name, '', Alignment, ['^sequence1', '^sequence2', '^[av].*'])) if os.path.exists(name) else None for name in input_files]
x_len = len(Fasta.alnToSeq(alignments[0][0].sequences[0]))
y_len = len(Fasta.alnToSeq(alignments[0][0].sequences[1]))
I = Image.new('RGB', (x_len * width + 50, y_len * width + 50), (255, 255, 255))
D = ImageDraw.Draw(I)
colors = [(0, 0, 0), (255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 0, 255), (0, 255, 255)]
i = -1
for aln in alignments:
i += 1
if aln == None:
continue
aln = list(aln)
if len(aln) == 0:
continue
aln = aln[0]
try:
annotation = aln.sequences[2]
coords = aln.getCoordPairs()
print coords
x_shift = width / 2 + 25 + i
y_shift = width / 2 + 25 + i * 2
D.line([(x * width + x_shift, y * width + y_shift) for
x, y, _ in coords], fill=colors[i])
if annotation != None:
for x, y, ind in coords:
if annotation[ind] != 'R':
continue
D.rectangle([(x * width + x_shift - width / 4, y * width + y_shift - width / 4),
(x * width + x_shift + width / 4, y * width + y_shift + width / 4)], outline=colors[i])
except IndexError:
pass
except IOError:
pass
del D
I.save(output_file)
def prepare_training_data(
self,
sequence_x,
annotations_x,
sequence_y,
annotations_y,
):
train_data = (list(), list(), list())
if self.insert_sequence == 0:
reference = sequence_x
annotations_r = annotations_x
space = sequence_y
annotations_s = annotations_y
else:
reference = sequence_y
annotations_r = annotations_y
space = sequence_x
annotations_s = annotations_x
sequence_rs = Fasta.alnToSeq(reference)
sequence_ss = Fasta.alnToSeq(space)
pos_s, pos_r = 0, 0
match_pos = set()
for i in range(len(space)):
br, bs = reference[i], space[i]
if bs != '-':
if br != '-':
match_pos.add((pos_r, pos_s))
pos_r += 1
pos_s += 1
continue
if br == '-':
continue
d = self.prepare_data(
sequence_rs,
pos_r,
annotations_r,
sequence_ss,
pos_s,
annotations_s,
0,
)
if d is not None:
train_data[0].append(d)
train_data[1].append(1)
train_data[2].append(1.0)
pos_r += 1
matches = sample(match_pos, len(train_data[0]))
for x, y in matches:
d = self.prepare_data(
sequence_rs,
x,
annotations_r,
sequence_ss,
y,
annotations_s,
0,
)
train_data[0].append(d)
train_data[1].append(0)
train_data[2].append(1.0)
return train_data
def prepare_training_data(
self,
sequence_x,
annotations_x,
sequence_y,
annotations_y,
):
train_data = (list(), list(), list())
if self.insert_sequence == 0:
reference = sequence_x
annotations_r = annotations_x
space = sequence_y
annotations_s = annotations_y
else:
reference = sequence_y
annotations_r = annotations_y
space = sequence_x
annotations_s = annotations_x
sequence_rs = Fasta.alnToSeq(reference)
sequence_ss = Fasta.alnToSeq(space)
pos_s, pos_r = 0, 0
match_pos = set()
for i in range(len(space)):
br, bs = reference[i], space[i]
if bs != '-':
if br != '-':
match_pos.add((pos_r, pos_s))
pos_r += 1
pos_s += 1
continue
if br == '-':
continue
d = self.prepare_data(
sequence_rs,
pos_r,
annotations_r,
sequence_ss,
pos_s,
annotations_s,
0,
)
if d is not None:
train_data[0].append(d)
train_data[1].append(1)
train_data[2].append(1.0)
pos_r += 1
matches = sample(match_pos, len(train_data[0]))
for x, y in matches:
d = self.prepare_data(
sequence_rs,
x,
annotations_r,
sequence_ss,
y,
annotations_s,
0,
)
train_data[0].append(d)
train_data[1].append(0)
train_data[2].append(1.0)
return train_data