def test_aligner_should_provide_Gotoh_algorithm(self):
score, A, B = aligner.align('ACACACTA', 'AGCACACA', 'GL', {
'match': 2,
'mismatch': -1,
'indel': -1
})
self.assertEqual(score, 12)
self.assertIn((A, B), [('A-CACACTA', 'AGCACAC-A')])
score, A, B = aligner.align('CGTGAATTCAT', 'GACTTAC', 'GL', {
'match': 5,
'mismatch': -3,
'indel': -4
})
self.assertEqual(score, 18)
self.assertIn((A, B), [('GAATTCA', 'GACTT-A'), ('GAATT-C', 'GACTTAC')])
score, A, B = aligner.align('CGGTCATAC', 'CGGAT', 'GG', {
'match': 1,
'mismatch': -1,
'indel': -1,
'gap_opening': -5
})
self.assertEqual(score, -5)
self.assertIn((A, B), [('CGGTCATAC', 'CGG----AT')])
def test_aligner_should_provide_Smith_Waterman_algorithm(self):
score, A, B = aligner.align('ACACACTA', 'AGCACACA', 'SW', {'match': 2, 'mismatch': -1, 'indel': -1})
self.assertEqual(score, 12)
self.assertIn((A, B), [('A-CACACTA', 'AGCACAC-A')])
score, A, B = aligner.align('CGTGAATTCAT', 'GACTTAC', 'SW', {'match': 5, 'mismatch': -3, 'indel': -4})
self.assertEqual(score, 18)
self.assertIn((A, B), [('GAATTCA', 'GACTT-A'), ('GAATT-C', 'GACTTAC')])
def test_aligner_should_accept_custom_penalties(self):
score, A, B = aligner.align('GCATGCU', 'GCATGCU', None, {'match': 10, 'mismatch': -1, 'indel': -1}) # 7 matches
self.assertEqual(score, 70)
score, A, B = aligner.align('GCATGCU', 'GTATGAG', None, {'match': 0, 'mismatch': -3, 'indel': -10}) # 3 mutations
self.assertEqual(score, -9)
score, A, B = aligner.align('GCTGCU', 'GCATGC', None, {'match': 0, 'mismatch': -10, 'indel': -4}) # 2 deletions
self.assertEqual(score, -8)
def test_aligner_should_provide_Gotoh_algorithm(self):
score, A, B = aligner.align('ACACACTA', 'AGCACACA', 'GL', {'match': 2, 'mismatch': -1, 'indel': -1})
self.assertEqual(score, 12)
self.assertIn((A, B), [('A-CACACTA', 'AGCACAC-A')])
score, A, B = aligner.align('CGTGAATTCAT', 'GACTTAC', 'GL', {'match': 5, 'mismatch': -3, 'indel': -4})
self.assertEqual(score, 18)
self.assertIn((A, B), [('GAATTCA', 'GACTT-A'), ('GAATT-C', 'GACTTAC')])
score, A, B = aligner.align('CGGTCATAC', 'CGGAT', 'GG', {'match': 1, 'mismatch': -1, 'indel': -1, 'gap_opening': -5})
self.assertEqual(score, -5)
self.assertIn((A, B), [('CGGTCATAC', 'CGG----AT')])
def test_aligner_should_work_on_Wikipedia_examples(self):
score, A, B = aligner.align('GCATGCU', 'GATTACA', None)
self.assertEqual(score, 0)
self.assertIn(
(A, B),
[('GCATG-CU', 'G-ATTACA'), ('GCA-TGCU', 'G-ATTACA'),
('GCAT-GCU', 'G-ATTACA')]) # Accept one of possible variants
def test_aligner_should_provide_Smith_Waterman_algorithm(self):
score, A, B = aligner.align('ACACACTA', 'AGCACACA', 'SW', {
'match': 2,
'mismatch': -1,
'indel': -1
})
self.assertEqual(score, 12)
self.assertIn((A, B), [('A-CACACTA', 'AGCACAC-A')])
score, A, B = aligner.align('CGTGAATTCAT', 'GACTTAC', 'SW', {
'match': 5,
'mismatch': -3,
'indel': -4
})
self.assertEqual(score, 18)
self.assertIn((A, B), [('GAATTCA', 'GACTT-A'), ('GAATT-C', 'GACTTAC')])
def test_aligner_should_detect_deletions(self, sequence_length=100):
alphabet = ['A', 'C', 'G', 'T']
A = self._generate_random_sequence(sequence_length, alphabet)
B = []
deletion_count = 0
for i in range(len(A)):
if np.random.random() < 0.2:
deletion_count = deletion_count + 1
else:
B.append(A[i])
score, _, _ = aligner.align(''.join(A), ''.join(A), None)
self.assertEqual(score, sequence_length)
score, _, _ = aligner.align(''.join(A), ''.join(B), None, {'match': 1, 'mismatch': -100, 'indel': -1})
self.assertEqual(score, sequence_length + deletion_count * -2)
def test_aligner_should_detect_insertions(self, sequence_length=100):
alphabet = ['A', 'C', 'G', 'T']
A = self._generate_random_sequence(sequence_length, alphabet)
B = []
insertion_count = 0
for i in range(len(A)):
B.append(A[i])
if np.random.random() < 0.2:
insertion_count = insertion_count + 1
B.append(alphabet[np.random.randint(len(alphabet))])
score, _, _ = aligner.align(''.join(A), ''.join(A), None)
self.assertEqual(score, sequence_length)
score, _, _ = aligner.align(''.join(A), ''.join(B), None, {'match': 1, 'mismatch': -100, 'indel': -1})
self.assertEqual(score, sequence_length + insertion_count + insertion_count * -2) # Insertions also increase total sequence length
def test_aligner_should_detect_mutations(self, sequence_length=100):
alphabet = ['A', 'C', 'G', 'T']
A = self._generate_random_sequence(sequence_length, alphabet)
B = list(A)
mutation_count = 0
for i in range(len(A)):
if np.random.random() < 0.2:
mutation = alphabet[np.random.randint(len(alphabet))]
if B[i] != mutation:
mutation_count = mutation_count + 1
B[i] = mutation
score, _, _ = aligner.align(''.join(A), ''.join(A), None)
self.assertEqual(score, sequence_length)
score, _, _ = aligner.align(''.join(A), ''.join(B), None, {'match': 1, 'mismatch': -1, 'indel': -10})
self.assertEqual(score, sequence_length + mutation_count * -2)
def test_aligner_should_accept_custom_penalties(self):
score, A, B = aligner.align('GCATGCU', 'GCATGCU', None, {
'match': 10,
'mismatch': -1,
'indel': -1
}) # 7 matches
self.assertEqual(score, 70)
score, A, B = aligner.align('GCATGCU', 'GTATGAG', None, {
'match': 0,
'mismatch': -3,
'indel': -10
}) # 3 mutations
self.assertEqual(score, -9)
score, A, B = aligner.align('GCTGCU', 'GCATGC', None, {
'match': 0,
'mismatch': -10,
'indel': -4
}) # 2 deletions
self.assertEqual(score, -8)
def test_aligner_should_detect_deletions(self, sequence_length=100):
alphabet = ['A', 'C', 'G', 'T']
A = self._generate_random_sequence(sequence_length, alphabet)
B = []
deletion_count = 0
for i in range(len(A)):
if np.random.random() < 0.2:
deletion_count = deletion_count + 1
else:
B.append(A[i])
score, _, _ = aligner.align(''.join(A), ''.join(A), None)
self.assertEqual(score, sequence_length)
score, _, _ = aligner.align(''.join(A), ''.join(B), None, {
'match': 1,
'mismatch': -100,
'indel': -1
})
self.assertEqual(score, sequence_length + deletion_count * -2)
def test_aligner_should_detect_mutations(self, sequence_length=100):
alphabet = ['A', 'C', 'G', 'T']
A = self._generate_random_sequence(sequence_length, alphabet)
B = list(A)
mutation_count = 0
for i in range(len(A)):
if np.random.random() < 0.2:
mutation = alphabet[np.random.randint(len(alphabet))]
if B[i] != mutation:
mutation_count = mutation_count + 1
B[i] = mutation
score, _, _ = aligner.align(''.join(A), ''.join(A), None)
self.assertEqual(score, sequence_length)
score, _, _ = aligner.align(''.join(A), ''.join(B), None, {
'match': 1,
'mismatch': -1,
'indel': -10
})
self.assertEqual(score, sequence_length + mutation_count * -2)
def test_aligner_should_detect_insertions(self, sequence_length=100):
alphabet = ['A', 'C', 'G', 'T']
A = self._generate_random_sequence(sequence_length, alphabet)
B = []
insertion_count = 0
for i in range(len(A)):
B.append(A[i])
if np.random.random() < 0.2:
insertion_count = insertion_count + 1
B.append(alphabet[np.random.randint(len(alphabet))])
score, _, _ = aligner.align(''.join(A), ''.join(A), None)
self.assertEqual(score, sequence_length)
score, _, _ = aligner.align(''.join(A), ''.join(B), None, {
'match': 1,
'mismatch': -100,
'indel': -1
})
self.assertEqual(score,
sequence_length + insertion_count + insertion_count *
-2) # Insertions also increase total sequence length
def test_aligner_should_raise_UnknownAlgorithmError_on_call_with_unknown_method(
self):
with self.assertRaises(aligner.UnknownAlgorithmError):
aligner.align('GCATGCU', 'GATTACA', method='XY')
def test_aligner_should_work_on_Wikipedia_examples(self):
score, A, B = aligner.align('GCATGCU', 'GATTACA', None)
self.assertEqual(score, 0)
self.assertIn((A, B), [('GCATG-CU', 'G-ATTACA'), ('GCA-TGCU', 'G-ATTACA'), ('GCAT-GCU', 'G-ATTACA')]) # Accept one of possible variants
def test_aligner_should_raise_UnknownAlgorithmError_on_call_with_unknown_method(self):
with self.assertRaises(aligner.UnknownAlgorithmError):
aligner.align('GCATGCU', 'GATTACA', method='XY')