def test_ProfileDistance_dynamic_smooth(self):
# If we use function=min and threshold=0, we should get the following
# transformation:
#
# | before | after
# ----------+------------------+-----------------
# | 0111111111111011 | 3000111111113000
# profile A | ACGTACGTACGTACGT | ACGTACGTACGTACGT
# | AAAACCCCGGGGTTTT | AAAACCCCGGGGTTTT
# ----------+------------------+-----------------
# | 0101111111111111 | 2000111111114000
# profile B | ACGTACGTACGTACGT | ACGTACGTACGTACGT
# | AAAACCCCGGGGTTTT | AAAACCCCGGGGTTTT
counts_a = Counter(['AC', 'AG', 'AT', 'CA', 'CC', 'CG', 'CT', 'GA', 'GC', 'GG', 'GT', 'TA', 'TG', 'TT'])
counts_b = Counter(['AC', 'AT', 'CA', 'CC', 'CG', 'CT', 'GA', 'GC', 'GG', 'GT', 'TA', 'TC', 'TG', 'TT'])
profile_a = klib.Profile(utils.as_array(counts_a, 2))
profile_b = klib.Profile(utils.as_array(counts_b, 2))
k_dist = kdistlib.ProfileDistance()
k_dist.dynamic_smooth(profile_a, profile_b)
counts_a = Counter(['AA', 'AA', 'AA', 'CA', 'CC', 'CG', 'CT', 'GA', 'GC', 'GG', 'GT', 'TA', 'TA', 'TA'])
counts_b = Counter(['AA', 'AA', 'CA', 'CC', 'CG', 'CT', 'GA', 'GC', 'GG', 'GT', 'TA', 'TA', 'TA', 'TA'])
np.testing.assert_array_equal(profile_a.counts, utils.as_array(counts_a, 2))
np.testing.assert_array_equal(profile_b.counts, utils.as_array(counts_b, 2))
def test_ProfileDistance_distance(self):
counts_a = Counter(['AC', 'AG', 'AT', 'CA', 'CC', 'CG', 'CT', 'GA', 'GC', 'GG', 'GT', 'TA', 'TG', 'TT'])
counts_b = Counter(['AC', 'AT', 'CA', 'CC', 'CG', 'CT', 'GA', 'GC', 'GG', 'GT', 'TA', 'TC', 'TG', 'TT'])
profile_a = klib.Profile(utils.as_array(counts_a, 2))
profile_b = klib.Profile(utils.as_array(counts_b, 2))
k_dist = kdistlib.ProfileDistance()
assert k_dist.distance(profile_a, profile_b) == 0.0625
def test_profile_merge(self):
counts_left = utils.counts(utils.SEQUENCES_LEFT, 8)
counts_right = utils.counts(utils.SEQUENCES_RIGHT, 8)
profile_left = klib.Profile(utils.as_array(counts_left, 8))
profile_right = klib.Profile(utils.as_array(counts_right, 8))
profile_left.merge(profile_right)
utils.test_profile(profile_left, counts_left + counts_right, 8)
def test_profile_merge(self):
counts_left = utils.counts(utils.SEQUENCES_LEFT, 8)
counts_right = utils.counts(utils.SEQUENCES_RIGHT, 8)
profile_left = klib.Profile(utils.as_array(counts_left, 8))
profile_right = klib.Profile(utils.as_array(counts_right, 8))
profile_left.merge(profile_right)
utils.test_profile(profile_left, counts_left + counts_right, 8)
def test_ProfileDistance_distance_k8(self):
counts_a = utils.counts(utils.SEQUENCES_LEFT, 8)
counts_b = utils.counts(utils.SEQUENCES_RIGHT, 8)
profile_a = klib.Profile(utils.as_array(counts_a, 8))
profile_b = klib.Profile(utils.as_array(counts_b, 8))
k_dist = kdistlib.ProfileDistance()
np.testing.assert_almost_equal(k_dist.distance(profile_a, profile_b), 0.4626209322)
def test_ProfileDistance_distance_k8(self):
counts_a = utils.counts(utils.SEQUENCES_LEFT, 8)
counts_b = utils.counts(utils.SEQUENCES_RIGHT, 8)
profile_a = klib.Profile(utils.as_array(counts_a, 8))
profile_b = klib.Profile(utils.as_array(counts_b, 8))
k_dist = kdistlib.ProfileDistance()
np.testing.assert_almost_equal(k_dist.distance(profile_a, profile_b),
0.4626209322)
def test_get_stats(self):
counts = utils.counts(utils.SEQUENCES, 8)
out = StringIO()
with utils.open_profile(self.profile(counts, 8)) as input_handle:
kmer.get_stats(input_handle, out)
name, mean, std = out.getvalue().strip().split()
assert name == '1'
assert mean == '%.10f' % np.mean(utils.as_array(counts, 8))
assert std == '%.10f' % np.std(utils.as_array(counts, 8))
def test_get_stats(self):
counts = utils.counts(utils.SEQUENCES, 8)
out = StringIO()
with utils.open_profile(self.profile(counts, 8)) as input_handle:
kmer.get_stats(input_handle, out)
name, mean, std = out.getvalue().strip().split()
assert name == '1'
assert mean == '%.10f' % np.mean(utils.as_array(counts, 8))
assert std == '%.10f' % np.std(utils.as_array(counts, 8))
def test_ProfileDistance_distance_unmodified(self):
counts_a = utils.counts(utils.SEQUENCES_LEFT, 8)
counts_b = utils.counts(utils.SEQUENCES_RIGHT, 8)
profile_a = klib.Profile(utils.as_array(counts_a, 8))
profile_b = klib.Profile(utils.as_array(counts_b, 8))
k_dist = kdistlib.ProfileDistance(do_balance=True)
k_dist.distance(profile_a, profile_b)
utils.test_profile(profile_a, counts_a, 8)
utils.test_profile(profile_b, counts_b, 8)
def test_distance_matrix_two(self):
counts_left = utils.counts(utils.SEQUENCES_LEFT, 8)
counts_right = utils.counts(utils.SEQUENCES_RIGHT, 8)
profiles = [klib.Profile(utils.as_array(counts_left, 8), 'a'),
klib.Profile(utils.as_array(counts_right, 8), 'b')]
k_dist = kdistlib.ProfileDistance()
out = StringIO()
kdistlib.distance_matrix(profiles, out, 2, k_dist)
assert out.getvalue().strip().split('\n') == ['2', 'a', 'b', '0.46']
def test_ProfileDistance_distance_unmodified(self):
counts_a = utils.counts(utils.SEQUENCES_LEFT, 8)
counts_b = utils.counts(utils.SEQUENCES_RIGHT, 8)
profile_a = klib.Profile(utils.as_array(counts_a, 8))
profile_b = klib.Profile(utils.as_array(counts_b, 8))
k_dist = kdistlib.ProfileDistance(do_balance=True)
k_dist.distance(profile_a, profile_b)
utils.test_profile(profile_a, counts_a, 8)
utils.test_profile(profile_b, counts_b, 8)
def test_distance_matrix_two(self):
counts_left = utils.counts(utils.SEQUENCES_LEFT, 8)
counts_right = utils.counts(utils.SEQUENCES_RIGHT, 8)
profiles = [
klib.Profile(utils.as_array(counts_left, 8), 'a'),
klib.Profile(utils.as_array(counts_right, 8), 'b')
]
k_dist = kdistlib.ProfileDistance()
out = StringIO()
kdistlib.distance_matrix(profiles, out, 2, k_dist)
assert out.getvalue().strip().split('\n') == ['2', 'a', 'b', '0.46']
def test_ProfileDistance_distance(self):
counts_a = Counter([
'AC', 'AG', 'AT', 'CA', 'CC', 'CG', 'CT', 'GA', 'GC', 'GG', 'GT',
'TA', 'TG', 'TT'
])
counts_b = Counter([
'AC', 'AT', 'CA', 'CC', 'CG', 'CT', 'GA', 'GC', 'GG', 'GT', 'TA',
'TC', 'TG', 'TT'
])
profile_a = klib.Profile(utils.as_array(counts_a, 2))
profile_b = klib.Profile(utils.as_array(counts_b, 2))
k_dist = kdistlib.ProfileDistance()
assert k_dist.distance(profile_a, profile_b) == 0.0625
def test_profile_reverse_complement_palindrome(self):
counts = utils.counts(['ACCTAGGT'], 8)
profile = klib.Profile(utils.as_array(counts, 8))
for i in range(profile.length):
assert (profile.binary_to_dna(profile.reverse_complement(i)) ==
utils.reverse_complement(profile.binary_to_dna(i)))
def _test_profile_split(self, sequences, length):
counts = utils.counts(sequences, length)
profile = klib.Profile(utils.as_array(counts, length))
left, right = profile.split()
assert len(left) == len(right)
assert sum(left) + sum(right) == sum(counts.values()) * 2
indices_left = {}
indices_right = {}
indices_palindrome = {}
for s, c in counts.items():
r = utils.reverse_complement(s)
if s < r:
indices_left[utils.count_index(s)] = c * 2
elif s > r:
indices_right[utils.count_index(r)] = counts[s] * 2
else:
indices_palindrome[utils.count_index(s)] = c
assert ([c for c in left if c > 0] == [
c for i, c in sorted(
list(indices_left.items()) + list(indices_palindrome.items()))
])
assert ([c for c in right if c > 0] == [
c for i, c in sorted(
list(indices_right.items()) + list(indices_palindrome.items()))
])
def test_profile_reverse_complement(self):
counts = utils.counts(utils.SEQUENCES, 8)
profile = klib.Profile(utils.as_array(counts, 8))
for i in range(profile.length):
assert (profile.binary_to_dna(profile.reverse_complement(i)) ==
utils.reverse_complement(profile.binary_to_dna(i)))
def _test_profile_split(self, sequences, length):
counts = utils.counts(sequences, length)
profile = klib.Profile(utils.as_array(counts, length))
left, right = profile.split()
assert len(left) == len(right)
assert sum(left) + sum(right) == sum(counts.values()) * 2
indices_left = {}
indices_right = {}
indices_palindrome = {}
for s, c in counts.items():
r = utils.reverse_complement(s)
if s < r:
indices_left[utils.count_index(s)] = c * 2
elif s > r:
indices_right[utils.count_index(r)] = counts[s] * 2
else:
indices_palindrome[utils.count_index(s)] = c
assert ([c for c in left if c > 0] ==
[c for i, c in sorted(list(indices_left.items()) +
list(indices_palindrome.items()))])
assert ([c for c in right if c > 0] ==
[c for i, c in sorted(list(indices_right.items()) +
list(indices_palindrome.items()))])
def test_profile_balance(self):
counts = utils.counts(utils.SEQUENCES, 8)
profile = klib.Profile(utils.as_array(counts, 8))
profile.balance()
counts.update(dict((utils.reverse_complement(s), c)
for s, c in counts.items()))
utils.test_profile(profile, counts, 8)
def test_profile_reverse_complement_palindrome(self):
counts = utils.counts(['ACCTAGGT'], 8)
profile = klib.Profile(utils.as_array(counts, 8))
for i in range(profile.length):
assert (profile.binary_to_dna(
profile.reverse_complement(i)) == utils.reverse_complement(
profile.binary_to_dna(i)))
def test_profile_reverse_complement(self):
counts = utils.counts(utils.SEQUENCES, 8)
profile = klib.Profile(utils.as_array(counts, 8))
for i in range(profile.length):
assert (profile.binary_to_dna(
profile.reverse_complement(i)) == utils.reverse_complement(
profile.binary_to_dna(i)))
def test_profile_balance_palindrome(self):
counts = utils.counts(['AATT'], 4)
profile = klib.Profile(utils.as_array(counts, 4))
profile.balance()
counts.update(
dict((utils.reverse_complement(s), c) for s, c in counts.items()))
utils.test_profile(profile, counts, 4)
def test_profile_balance(self):
counts = utils.counts(utils.SEQUENCES, 8)
profile = klib.Profile(utils.as_array(counts, 8))
profile.balance()
counts.update(
dict((utils.reverse_complement(s), c) for s, c in counts.items()))
utils.test_profile(profile, counts, 8)
def test_profile_balance_palindrome(self):
counts = utils.counts(['AATT'], 4)
profile = klib.Profile(utils.as_array(counts, 4))
profile.balance()
counts.update(dict((utils.reverse_complement(s), c)
for s, c in counts.items()))
utils.test_profile(profile, counts, 4)
def test_profile_print_counts(self, capsys):
counts = utils.counts(utils.SEQUENCES, 4)
profile = klib.Profile(utils.as_array(counts, 4))
profile.print_counts()
out, err = capsys.readouterr()
assert out == ''.join('%s %d\n' % (''.join(s), counts[''.join(s)])
for s in itertools.product('ACGT', repeat=4))
def test_profile_print_counts(self, capsys):
counts = utils.counts(utils.SEQUENCES, 4)
profile = klib.Profile(utils.as_array(counts, 4))
profile.print_counts()
out, err = capsys.readouterr()
assert out == ''.join('%s %d\n' % (''.join(s), counts[''.join(s)])
for s in itertools.product('ACGT', repeat=4))
def test_profile_shrink_max(self):
counts = utils.counts(utils.SEQUENCES, 4)
profile = klib.Profile(utils.as_array(counts, 4))
profile.shrink(3)
counts = Counter(
dict((t, sum(counts[u] for u in counts if u.startswith(t)))
for t in set(s[:-3] for s in counts)))
utils.test_profile(profile, counts, 1)
def test_profile_shrink_max(self):
counts = utils.counts(utils.SEQUENCES, 4)
profile = klib.Profile(utils.as_array(counts, 4))
profile.shrink(3)
counts = Counter(dict((t, sum(counts[u] for u in counts
if u.startswith(t)))
for t in set(s[:-3] for s in counts)))
utils.test_profile(profile, counts, 1)
def test_profile_save(self):
counts = utils.counts(utils.SEQUENCES, 4)
profile = klib.Profile(utils.as_array(counts, 4))
filename = self.empty()
with utils.open_profile(filename, 'w') as profile_handle:
profile.save(profile_handle)
utils.test_profile_file(filename, counts, 4)
def test_profile_save(self):
counts = utils.counts(utils.SEQUENCES, 4)
profile = klib.Profile(utils.as_array(counts, 4))
filename = self.empty()
with utils.open_profile(filename, 'w') as profile_handle:
profile.save(profile_handle)
utils.test_profile_file(filename, counts, 4)
def test_profile_shuffle(self):
counts = utils.counts(utils.SEQUENCES, 2)
profile = klib.Profile(utils.as_array(counts, 2))
np.random.seed(100)
profile.shuffle()
counts = dict(zip([''.join(s) for s in itertools.product('ACGT', repeat=2)],
[13, 7, 6, 18, 12, 1, 13, 17, 16, 12, 23, 27, 24, 17, 18, 12]))
utils.test_profile(profile, counts, 2)
def test_distance_matrix_one(self):
counts = utils.counts(utils.SEQUENCES, 8)
profiles = [klib.Profile(utils.as_array(counts, 8), 'a')]
k_dist = kdistlib.ProfileDistance()
out = StringIO()
kdistlib.distance_matrix(profiles, out, 2, k_dist)
assert out.getvalue().strip().split('\n') == ['1', 'a']
def test_distribution(self):
counts = utils.counts(utils.SEQUENCES, 8)
out = StringIO()
with utils.open_profile(self.profile(counts, 8)) as input_handle:
kmer.distribution(input_handle, out)
counter = Counter(utils.as_array(counts, 8))
assert out.getvalue() == '\n'.join('1 %i %i' % x
for x in sorted(counter.items())) + '\n'
def test_distribution(self):
counts = utils.counts(utils.SEQUENCES, 8)
out = StringIO()
with utils.open_profile(self.profile(counts, 8)) as input_handle:
kmer.distribution(input_handle, out)
counter = Counter(utils.as_array(counts, 8))
assert out.getvalue() == '\n'.join('1 %i %i' % x
for x in sorted(counter.items())) + '\n'
def test_distance_matrix_one(self):
counts = utils.counts(utils.SEQUENCES, 8)
profiles = [klib.Profile(utils.as_array(counts, 8), 'a')]
k_dist = kdistlib.ProfileDistance()
out = StringIO()
kdistlib.distance_matrix(profiles, out, 2, k_dist)
assert out.getvalue().strip().split('\n') == ['1', 'a']
def test_profile_shuffle(self):
counts = utils.counts(utils.SEQUENCES, 2)
profile = klib.Profile(utils.as_array(counts, 2))
np.random.seed(100)
profile.shuffle()
counts = dict(
zip([''.join(s) for s in itertools.product('ACGT', repeat=2)],
[13, 7, 6, 18, 12, 1, 13, 17, 16, 12, 23, 27, 24, 17, 18, 12]))
utils.test_profile(profile, counts, 2)
def test_ProfileDistance_dynamic_smooth(self):
# If we use function=min and threshold=0, we should get the following
# transformation:
#
# | before | after
# ----------+------------------+-----------------
# | 0111111111111011 | 3000111111113000
# profile A | ACGTACGTACGTACGT | ACGTACGTACGTACGT
# | AAAACCCCGGGGTTTT | AAAACCCCGGGGTTTT
# ----------+------------------+-----------------
# | 0101111111111111 | 2000111111114000
# profile B | ACGTACGTACGTACGT | ACGTACGTACGTACGT
# | AAAACCCCGGGGTTTT | AAAACCCCGGGGTTTT
counts_a = Counter([
'AC', 'AG', 'AT', 'CA', 'CC', 'CG', 'CT', 'GA', 'GC', 'GG', 'GT',
'TA', 'TG', 'TT'
])
counts_b = Counter([
'AC', 'AT', 'CA', 'CC', 'CG', 'CT', 'GA', 'GC', 'GG', 'GT', 'TA',
'TC', 'TG', 'TT'
])
profile_a = klib.Profile(utils.as_array(counts_a, 2))
profile_b = klib.Profile(utils.as_array(counts_b, 2))
k_dist = kdistlib.ProfileDistance()
k_dist.dynamic_smooth(profile_a, profile_b)
counts_a = Counter([
'AA', 'AA', 'AA', 'CA', 'CC', 'CG', 'CT', 'GA', 'GC', 'GG', 'GT',
'TA', 'TA', 'TA'
])
counts_b = Counter([
'AA', 'AA', 'CA', 'CC', 'CG', 'CT', 'GA', 'GC', 'GG', 'GT', 'TA',
'TA', 'TA', 'TA'
])
np.testing.assert_array_equal(profile_a.counts,
utils.as_array(counts_a, 2))
np.testing.assert_array_equal(profile_b.counts,
utils.as_array(counts_b, 2))
def test_profile_ratios_matrix(self):
counts = utils.counts(utils.SEQUENCES, 4)
profile = klib.Profile(utils.as_array(counts, 4))
ratios = profile._ratios_matrix()
total = sum(counts.values())
for left in itertools.product('ACGT', repeat=4):
for right in itertools.product('ACGT', repeat=4):
left = ''.join(left)
right = ''.join(right)
ratio = ratios[utils.count_index(left)][utils.count_index(right)]
try:
assert ratio == counts[left] / counts[right] / total
except ZeroDivisionError:
assert ratio == -1.0
def test_profile_freq_diff_matrix(self):
counts = utils.counts(utils.SEQUENCES, 4)
profile = klib.Profile(utils.as_array(counts, 4))
freq_diffs = profile._freq_diff_matrix()
total = sum(counts.values())
for left in itertools.product('ACGT', repeat=4):
for right in itertools.product('ACGT', repeat=4):
left = ''.join(left)
right = ''.join(right)
freq_diff = freq_diffs[utils.count_index(left)][utils.count_index(right)]
if counts[right] > 0:
assert freq_diff == abs(counts[left] - counts[right]) / total
else:
assert freq_diff == 0
def test_profile_ratios_matrix(self):
counts = utils.counts(utils.SEQUENCES, 4)
profile = klib.Profile(utils.as_array(counts, 4))
ratios = profile._ratios_matrix()
total = sum(counts.values())
for left in itertools.product('ACGT', repeat=4):
for right in itertools.product('ACGT', repeat=4):
left = ''.join(left)
right = ''.join(right)
ratio = ratios[utils.count_index(left)][utils.count_index(
right)]
try:
assert ratio == counts[left] / counts[right] / total
except ZeroDivisionError:
assert ratio == -1.0
def test_profile_freq_diff_matrix(self):
counts = utils.counts(utils.SEQUENCES, 4)
profile = klib.Profile(utils.as_array(counts, 4))
freq_diffs = profile._freq_diff_matrix()
total = sum(counts.values())
for left in itertools.product('ACGT', repeat=4):
for right in itertools.product('ACGT', repeat=4):
left = ''.join(left)
right = ''.join(right)
freq_diff = freq_diffs[utils.count_index(left)][
utils.count_index(right)]
if counts[right] > 0:
assert freq_diff == abs(counts[left] -
counts[right]) / total
else:
assert freq_diff == 0
def test_profile(self):
counts = utils.counts(utils.SEQUENCES, 8)
profile = klib.Profile(utils.as_array(counts, 8))
utils.test_profile(profile, counts, 8)
def test_profile(self):
counts = utils.counts(utils.SEQUENCES, 8)
profile = klib.Profile(utils.as_array(counts, 8))
utils.test_profile(profile, counts, 8)
def test_profile_binary_to_dna(self):
counts = utils.counts(utils.SEQUENCES, 4)
profile = klib.Profile(utils.as_array(counts, 4))
for i, s in enumerate(itertools.product('ACGT', repeat=4)):
assert ''.join(s) == profile.binary_to_dna(i)
def test_profile_shrink_invalid(self):
counts = utils.counts(utils.SEQUENCES, 4)
profile = klib.Profile(utils.as_array(counts, 4))
with pytest.raises(ValueError):
profile.shrink(4)
def test_profile_shrink_invalid(self):
counts = utils.counts(utils.SEQUENCES, 4)
profile = klib.Profile(utils.as_array(counts, 4))
with pytest.raises(ValueError):
profile.shrink(4)
def test_profile_binary_to_dna(self):
counts = utils.counts(utils.SEQUENCES, 4)
profile = klib.Profile(utils.as_array(counts, 4))
for i, s in enumerate(itertools.product('ACGT', repeat=4)):
assert ''.join(s) == profile.binary_to_dna(i)
def test_profile_name_with_slash(self):
counts = utils.counts(utils.SEQUENCES, 4)
with pytest.raises(ValueError):
klib.Profile(utils.as_array(counts, 4), name='abc/def')
def test_profile_name_with_slash(self):
counts = utils.counts(utils.SEQUENCES, 4)
with pytest.raises(ValueError):
klib.Profile(utils.as_array(counts, 4), name='abc/def')