def test_compute_score_and_traceback_matrices(self):
# these results were computed manually
expected_score_m = [[0, -5, -7, -9],
[-5, 2, -3, -5],
[-7, -3, 4, -1],
[-9, -5, -1, 6],
[-11, -7, -3, 1]]
expected_tback_m = [[0, 3, 3, 3],
[2, 1, 3, 3],
[2, 2, 1, 3],
[2, 2, 2, 1],
[2, 2, 2, 2]]
m = _make_nt_substitution_matrix(2, -1)
actual_score_m, actual_tback_m = _compute_score_and_traceback_matrices(
'ACG', 'ACGT', 5, 2, m)
np.testing.assert_array_equal(actual_score_m, expected_score_m)
np.testing.assert_array_equal(actual_tback_m, expected_tback_m)
# different sequences
# these results were computed manually
expected_score_m = [[0, -5, -7, -9],
[-5, 2, -3, -5],
[-7, -3, 4, -1],
[-9, -5, -1, 3],
[-11, -7, -3, -2]]
expected_tback_m = [[0, 3, 3, 3],
[2, 1, 3, 3],
[2, 2, 1, 3],
[2, 2, 2, 1],
[2, 2, 2, 1]]
m = _make_nt_substitution_matrix(2, -1)
actual_score_m, actual_tback_m = _compute_score_and_traceback_matrices(
'ACC', 'ACGT', 5, 2, m)
np.testing.assert_array_equal(actual_score_m, expected_score_m)
np.testing.assert_array_equal(actual_tback_m, expected_tback_m)
def test_make_nt_substitution_matrix(self):
expected = {'A': {'A': 1, 'C': -2, 'G': -2, 'T': -2},
'C': {'A': -2, 'C': 1, 'G': -2, 'T': -2},
'G': {'A': -2, 'C': -2, 'G': 1, 'T': -2},
'T': {'A': -2, 'C': -2, 'G': -2, 'T': 1}}
self.assertEqual(_make_nt_substitution_matrix(1, -2), expected)
expected = {'A': {'A': 5, 'C': -4, 'G': -4, 'T': -4},
'C': {'A': -4, 'C': 5, 'G': -4, 'T': -4},
'G': {'A': -4, 'C': -4, 'G': 5, 'T': -4},
'T': {'A': -4, 'C': -4, 'G': -4, 'T': 5}}
self.assertEqual(_make_nt_substitution_matrix(5, -4), expected)
def test_nucleotide_aligners_use_substitution_matrices(self):
alt_sub = _make_nt_substitution_matrix(10, -10)
# alternate substitution matrix yields different alignment (the
# aligned sequences and the scores are different) with local alignment
with warnings.catch_warnings():
warnings.simplefilter("ignore")
actual_no_sub = local_pairwise_align_nucleotide(
"GACCTTGACCAGGTACC", "GAACTTTGACGTAAC", gap_open_penalty=10.,
gap_extend_penalty=5., match_score=5, mismatch_score=-4)
actual_alt_sub = local_pairwise_align_nucleotide(
"GACCTTGACCAGGTACC", "GAACTTTGACGTAAC", gap_open_penalty=10.,
gap_extend_penalty=5., match_score=5, mismatch_score=-4,
substitution_matrix=alt_sub)
self.assertNotEqual(str(actual_no_sub[0]), str(actual_alt_sub[0]))
self.assertNotEqual(str(actual_no_sub[1]), str(actual_alt_sub[1]))
self.assertNotEqual(actual_no_sub.score(),
actual_alt_sub.score())
# alternate substitution matrix yields different alignment (the
# aligned sequences and the scores are different) with global alignment
with warnings.catch_warnings():
warnings.simplefilter("ignore")
actual_no_sub = local_pairwise_align_nucleotide(
"GACCTTGACCAGGTACC", "GAACTTTGACGTAAC", gap_open_penalty=10.,
gap_extend_penalty=5., match_score=5, mismatch_score=-4)
actual_alt_sub = global_pairwise_align_nucleotide(
"GACCTTGACCAGGTACC", "GAACTTTGACGTAAC", gap_open_penalty=10.,
gap_extend_penalty=5., match_score=5, mismatch_score=-4,
substitution_matrix=alt_sub)
self.assertNotEqual(str(actual_no_sub[0]), str(actual_alt_sub[0]))
self.assertNotEqual(str(actual_no_sub[1]), str(actual_alt_sub[1]))
self.assertNotEqual(actual_no_sub.score(),
actual_alt_sub.score())
def test_local_pairwise_align_nucleotide(self):
m = _make_nt_substitution_matrix(5, -4)
expected = ("ACCTTGACCAGGTACC", "ACTTTGAC---GTAAC", 41.0, 1, 2)
with warnings.catch_warnings():
warnings.simplefilter("ignore")
actual = local_pairwise_align_nucleotide(
"GACCTTGACCAGGTACC", "GAACTTTGACGTAAC", gap_open_penalty=5.,
gap_extend_penalty=0.5, match_score=5, mismatch_score=-4)
self.assertEqual(str(actual[0]), expected[0])
self.assertEqual(str(actual[1]), expected[1])
self.assertEqual(actual.score(), expected[2])
self.assertEqual(actual.start_end_positions(), [(1, 16), (2, 14)])
self.assertEqual(actual.ids(), list('01'))
expected = ("ACCTTGAC", "ACTTTGAC", 31.0, 1, 2)
with warnings.catch_warnings():
warnings.simplefilter("ignore")
actual = local_pairwise_align_nucleotide(
"GACCTTGACCAGGTACC", "GAACTTTGACGTAAC", gap_open_penalty=10.,
gap_extend_penalty=5., match_score=5, mismatch_score=-4)
self.assertEqual(str(actual[0]), expected[0])
self.assertEqual(str(actual[1]), expected[1])
self.assertEqual(actual.score(), expected[2])
self.assertEqual(actual.start_end_positions(), [(1, 8), (2, 9)])
self.assertEqual(actual.ids(), list('01'))
# DNA (rather than str) as input
expected = ("ACCTTGAC", "ACTTTGAC", 31.0, 1, 2)
with warnings.catch_warnings():
warnings.simplefilter("ignore")
actual = local_pairwise_align_nucleotide(
DNA("GACCTTGACCAGGTACC", "s1"), DNA("GAACTTTGACGTAAC", "s2"),
gap_open_penalty=10., gap_extend_penalty=5., match_score=5,
mismatch_score=-4)
self.assertEqual(str(actual[0]), expected[0])
self.assertEqual(str(actual[1]), expected[1])
self.assertEqual(actual.score(), expected[2])
self.assertEqual(actual.start_end_positions(), [(1, 8), (2, 9)])
self.assertEqual(actual.ids(), ["s1", "s2"])
# ids are provided if they're not passed in
with warnings.catch_warnings():
warnings.simplefilter("ignore")
actual = local_pairwise_align_nucleotide(
DNA("GACCTTGACCAGGTACC"), DNA("GAACTTTGACGTAAC"),
gap_open_penalty=10., gap_extend_penalty=5., match_score=5,
mismatch_score=-4)
self.assertEqual(actual.ids(), list('01'))
def test_compute_score_and_traceback_matrices_invalid(self):
# if the sequence contains a character that is not in the
# substitution matrix, an informative error should be raised
m = _make_nt_substitution_matrix(2, -1)
self.assertRaises(ValueError, _compute_score_and_traceback_matrices,
'AWG', 'ACGT', 5, 2, m)
