def test_needleman_wunsch_dist_abs_nialls(self):
"""Test abydos.distance.NeedlemanWunsch.dist_abs (Nialls set)."""
# checked against http://ds9a.nl/nwunsch/ (mismatch=1, gap=2, skew=2)
nw_vals = (5, 0, -2, 3, 1, 1, -2, -2, -1, -3, -3, -5, -3, -7, -7, -19)
nw2 = NeedlemanWunsch(2, _sim_nw)
for i in range(len(NIALL)):
self.assertEqual(nw2.sim_score(NIALL[0], NIALL[i]), nw_vals[i])
def test_needleman_wunsch_dist_abs_nialls(self):
"""Test abydos.distance.NeedlemanWunsch.dist_abs (Nialls set)."""
# checked against http://ds9a.nl/nwunsch/ (mismatch=1, gap=2, skew=2)
nw_vals = (5, 0, -2, 3, 1, 1, -2, -2, -1, -3, -3, -5, -3, -7, -7, -19)
nw2 = NeedlemanWunsch(2, _sim_nw)
for i in range(len(NIALL)):
self.assertEqual(nw2.dist_abs(NIALL[0], NIALL[i]), nw_vals[i])
def test_needleman_wunsch_dist_abs(self):
"""Test abydos.distance.NeedlemanWunsch.dist_abs."""
self.assertEqual(NeedlemanWunsch().dist_abs('', ''), 0)
# https://en.wikipedia.org/wiki/Needleman–Wunsch_algorithm
self.assertEqual(
NeedlemanWunsch(1, _sim_nw).dist_abs('GATTACA', 'GCATGCU'), 0
)
self.assertEqual(
NeedlemanWunsch(5, _sim_wikipedia).dist_abs(
'AGACTAGTTAC', 'CGAGACGT'
),
16,
)
# checked against http://ds9a.nl/nwunsch/ (mismatch=1, gap=5, skew=5)
nw5 = NeedlemanWunsch(5, _sim_nw)
self.assertEqual(nw5.dist_abs('CGATATCAG', 'TGACGSTGC'), -5)
self.assertEqual(nw5.dist_abs('AGACTAGTTAC', 'TGACGSTGC'), -7)
self.assertEqual(nw5.dist_abs('AGACTAGTTAC', 'CGAGACGT'), -15)
# test wrapper
self.assertEqual(
needleman_wunsch('AGACTAGTTAC', 'CGAGACGT', 5, _sim_nw), -15
)
def test_gotoh_sim_score(self):
"""Test abydos.distance.Gotoh.sim_score."""
self.assertEqual(Gotoh().sim_score('', ''), 0)
# https://en.wikipedia.org/wiki/Needleman–Wunsch_algorithm
self.assertEqual(
Gotoh(1, 1, _sim_nw).sim_score('GATTACA', 'GCATGCU'), 0)
self.assertGreaterEqual(
Gotoh(1, 0.5, _sim_nw).sim_score('GATTACA', 'GCATGCU'),
NeedlemanWunsch(1, _sim_nw).sim_score('GATTACA', 'GCATGCU'),
)
self.assertEqual(
Gotoh(5, 5, _sim_wikipedia).sim_score('AGACTAGTTAC', 'CGAGACGT'),
16,
)
self.assertGreaterEqual(
Gotoh(5, 2, _sim_wikipedia).sim_score('AGACTAGTTAC', 'CGAGACGT'),
NeedlemanWunsch(5,
_sim_wikipedia).sim_score('AGACTAGTTAC',
'CGAGACGT'),
)
# checked against http://ds9a.nl/nwunsch/ (mismatch=1, gap=5, skew=5)
self.assertEqual(
Gotoh(5, 5, _sim_nw).sim_score('CGATATCAG', 'TGACGSTGC'), -5)
self.assertGreaterEqual(
Gotoh(5, 2, _sim_nw).sim_score('CGATATCAG', 'TGACGSTGC'),
NeedlemanWunsch(5, _sim_nw).sim_score('CGATATCAG', 'TGACGSTGC'),
)
self.assertEqual(
Gotoh(5, 5, _sim_nw).sim_score('AGACTAGTTAC', 'TGACGSTGC'), -7)
self.assertGreaterEqual(
Gotoh(5, 2, _sim_nw).sim_score('AGACTAGTTAC', 'TGACGSTGC'),
NeedlemanWunsch(5, _sim_nw).sim_score('AGACTAGTTAC', 'TGACGSTGC'),
)
self.assertEqual(
Gotoh(5, 5, _sim_nw).sim_score('AGACTAGTTAC', 'CGAGACGT'), -15)
self.assertGreaterEqual(
Gotoh(5, 2, _sim_nw).sim_score('AGACTAGTTAC', 'CGAGACGT'),
NeedlemanWunsch(5, _sim_nw).sim_score('AGACTAGTTAC', 'CGAGACGT'),
)
# Test cases where one of the strings has length 1
self.assertAlmostEqual(Gotoh().sim_score('', 'a'), -1.0)
self.assertAlmostEqual(Gotoh().sim_score('abc', 'a'), -0.4)
self.assertAlmostEqual(Gotoh().sim_score('a', ''), -1.0)
self.assertAlmostEqual(Gotoh().sim_score('a', 'abc'), -0.4)
# Test wrapper
self.assertEqual(
gotoh('AGACTAGTTAC', 'CGAGACGT', 5, 5, _sim_wikipedia), 16)
def test_needleman_wunsch_sim_score(self):
"""Test abydos.distance.NeedlemanWunsch.sim_score."""
self.assertEqual(NeedlemanWunsch().sim_score('', ''), 0)
# https://en.wikipedia.org/wiki/Needleman–Wunsch_algorithm
self.assertEqual(
NeedlemanWunsch(1, _sim_nw).sim_score('GATTACA', 'GCATGCU'), 0
)
self.assertEqual(
NeedlemanWunsch(5, _sim_wikipedia).sim_score(
'AGACTAGTTAC', 'CGAGACGT'
),
16,
)
# checked against http://ds9a.nl/nwunsch/ (mismatch=1, gap=5, skew=5)
nw5 = NeedlemanWunsch(5, _sim_nw)
self.assertEqual(nw5.sim_score('CGATATCAG', 'TGACGSTGC'), -5)
self.assertEqual(nw5.sim_score('AGACTAGTTAC', 'TGACGSTGC'), -7)
self.assertEqual(nw5.sim_score('AGACTAGTTAC', 'CGAGACGT'), -15)
# test wrapper
self.assertEqual(
needleman_wunsch('AGACTAGTTAC', 'CGAGACGT', 5, _sim_nw), -15
)
def _sim_wikipedia(src, tar):
"""Return a similarity score for two DNA base pairs.
Values copied from:
https://en.wikipedia.org/wiki/Needleman%E2%80%93Wunsch_algorithm
Parameters
----------
src : str
Source string for comparison
tar : str
Target string for comparison
Returns
-------
int: similarity of two DNA base pairs
"""
nw_matrix = {
('A', 'A'): 10,
('G', 'G'): 7,
('C', 'C'): 9,
('T', 'T'): 8,
('A', 'G'): -1,
('A', 'C'): -3,
('A', 'T'): -4,
('G', 'C'): -5,
('G', 'T'): -3,
('C', 'T'): 0,
}
return NeedlemanWunsch.sim_matrix(
src, tar, nw_matrix, symmetric=True, alphabet='CGAT'
)
def _sim_wikipedia(src, tar):
"""Return a similarity score for two DNA base pairs.
Values copied from:
https://en.wikipedia.org/wiki/Needleman%E2%80%93Wunsch_algorithm
Parameters
----------
src : str
Source string for comparison
tar : str
Target string for comparison
Returns
-------
int: similarity of two DNA base pairs
"""
nw_matrix = {
('A', 'A'): 10,
('G', 'G'): 7,
('C', 'C'): 9,
('T', 'T'): 8,
('A', 'G'): -1,
('A', 'C'): -3,
('A', 'T'): -4,
('G', 'C'): -5,
('G', 'T'): -3,
('C', 'T'): 0,
}
return NeedlemanWunsch.sim_matrix(
src, tar, nw_matrix, symmetric=True, alphabet='CGAT'
)
def test_gotoh_sim_score_nialls(self):
"""Test abydos.distance.Gotoh.sim_score (Nialls set)."""
# checked against http://ds9a.nl/nwunsch/ (mismatch=1, gap=2, skew=2)
nw_vals = (5, 0, -2, 3, 1, 1, -2, -2, -1, -3, -3, -5, -3, -7, -7, -19)
g22 = Gotoh(2, 2, _sim_nw)
for i in range(len(NIALL)):
self.assertEqual(g22.sim_score(NIALL[0], NIALL[i]), nw_vals[i])
nw_vals2 = (5, 0, -2, 3, 1, 1, -2, -2, -1, -2, -3, -3, -2, -6, -6, -8)
g21 = Gotoh(2, 1, _sim_nw)
g205 = Gotoh(2, 0.5, _sim_nw)
nw2 = NeedlemanWunsch(2, _sim_nw)
for i in range(len(NIALL)):
self.assertEqual(g21.sim_score(NIALL[0], NIALL[i]), nw_vals2[i])
self.assertGreaterEqual(
g205.sim_score(NIALL[0], NIALL[i]),
nw2.sim_score(NIALL[0], NIALL[i]),
)
def test_gotoh_dist_abs_nialls(self):
"""Test abydos.distance.Gotoh.dist_abs (Nialls set)."""
# checked against http://ds9a.nl/nwunsch/ (mismatch=1, gap=2, skew=2)
nw_vals = (5, 0, -2, 3, 1, 1, -2, -2, -1, -3, -3, -5, -3, -7, -7, -19)
g22 = Gotoh(2, 2, _sim_nw)
for i in range(len(NIALL)):
self.assertEqual(g22.dist_abs(NIALL[0], NIALL[i]), nw_vals[i])
nw_vals2 = (5, 0, -2, 3, 1, 1, -2, -2, -1, -2, -3, -3, -2, -6, -6, -8)
g21 = Gotoh(2, 1, _sim_nw)
g205 = Gotoh(2, 0.5, _sim_nw)
nw2 = NeedlemanWunsch(2, _sim_nw)
for i in range(len(NIALL)):
self.assertEqual(g21.dist_abs(NIALL[0], NIALL[i]), nw_vals2[i])
self.assertGreaterEqual(
g205.dist_abs(NIALL[0], NIALL[i]),
nw2.dist_abs(NIALL[0], NIALL[i]),
)
def test_sim_matrix(self):
"""Test abydos.distance.NeedlemanWunsch.sim_matrix."""
self.assertEqual(NeedlemanWunsch.sim_matrix('', ''), 1)
self.assertEqual(NeedlemanWunsch.sim_matrix('', 'a'), 0)
self.assertEqual(NeedlemanWunsch.sim_matrix('a', ''), 0)
self.assertEqual(NeedlemanWunsch.sim_matrix('a', 'a'), 1)
self.assertEqual(NeedlemanWunsch.sim_matrix('abcd', 'abcd'), 1)
self.assertEqual(NeedlemanWunsch.sim_matrix('abcd', 'dcba'), 0)
self.assertEqual(NeedlemanWunsch.sim_matrix('abc', 'cba'), 0)
# https://en.wikipedia.org/wiki/Needleman–Wunsch_algorithm
self.assertEqual(_sim_wikipedia('A', 'C'), -3)
self.assertEqual(_sim_wikipedia('G', 'G'), 7)
self.assertEqual(_sim_wikipedia('A', 'A'), 10)
self.assertEqual(_sim_wikipedia('T', 'A'), -4)
self.assertEqual(_sim_wikipedia('T', 'C'), 0)
self.assertEqual(_sim_wikipedia('A', 'G'), -1)
self.assertEqual(_sim_wikipedia('C', 'T'), 0)
self.assertRaises(
ValueError, NeedlemanWunsch.sim_matrix, 'abc', 'cba', alphabet='ab'
)
self.assertRaises(
ValueError, NeedlemanWunsch.sim_matrix, 'abc', 'ba', alphabet='ab'
)
self.assertRaises(
ValueError, NeedlemanWunsch.sim_matrix, 'ab', 'cba', alphabet='ab'
)
def test_sim_matrix(self):
"""Test abydos.distance.NeedlemanWunsch.sim_matrix."""
self.assertEqual(NeedlemanWunsch.sim_matrix('', ''), 1)
self.assertEqual(NeedlemanWunsch.sim_matrix('', 'a'), 0)
self.assertEqual(NeedlemanWunsch.sim_matrix('a', ''), 0)
self.assertEqual(NeedlemanWunsch.sim_matrix('a', 'a'), 1)
self.assertEqual(NeedlemanWunsch.sim_matrix('abcd', 'abcd'), 1)
self.assertEqual(NeedlemanWunsch.sim_matrix('abcd', 'dcba'), 0)
self.assertEqual(NeedlemanWunsch.sim_matrix('abc', 'cba'), 0)
# https://en.wikipedia.org/wiki/Needleman–Wunsch_algorithm
self.assertEqual(_sim_wikipedia('A', 'C'), -3)
self.assertEqual(_sim_wikipedia('G', 'G'), 7)
self.assertEqual(_sim_wikipedia('A', 'A'), 10)
self.assertEqual(_sim_wikipedia('T', 'A'), -4)
self.assertEqual(_sim_wikipedia('T', 'C'), 0)
self.assertEqual(_sim_wikipedia('A', 'G'), -1)
self.assertEqual(_sim_wikipedia('C', 'T'), 0)
self.assertRaises(
ValueError, NeedlemanWunsch.sim_matrix, 'abc', 'cba', alphabet='ab'
)
self.assertRaises(
ValueError, NeedlemanWunsch.sim_matrix, 'abc', 'ba', alphabet='ab'
)
self.assertRaises(
ValueError, NeedlemanWunsch.sim_matrix, 'ab', 'cba', alphabet='ab'
)
def test_gotoh_sim(self):
"""Test abydos.distance.Gotoh.sim."""
self.assertEqual(Gotoh().sim('', ''), 1.0)
# https://en.wikipedia.org/wiki/Needleman–Wunsch_algorithm
self.assertEqual(Gotoh(1, 1, _sim_nw).sim('GATTACA', 'GCATGCU'), 0)
self.assertGreaterEqual(
Gotoh(1, 0.5, _sim_nw).sim('GATTACA', 'GCATGCU'),
NeedlemanWunsch(1, _sim_nw).sim('GATTACA', 'GCATGCU'),
)
self.assertEqual(
Gotoh(5, 5, _sim_wikipedia).sim('AGACTAGTTAC', 'CGAGACGT'),
0.19950186722152657,
)
self.assertGreaterEqual(
Gotoh(5, 2, _sim_wikipedia).sim('AGACTAGTTAC', 'CGAGACGT'),
NeedlemanWunsch(5, _sim_wikipedia).sim('AGACTAGTTAC', 'CGAGACGT'),
)
def test_gotoh_dist_abs(self):
"""Test abydos.distance.Gotoh.dist_abs."""
self.assertEqual(Gotoh().dist_abs('', ''), 0)
# https://en.wikipedia.org/wiki/Needleman–Wunsch_algorithm
self.assertEqual(
Gotoh(1, 1, _sim_nw).dist_abs('GATTACA', 'GCATGCU'), 0)
self.assertGreaterEqual(
Gotoh(1, 0.5, _sim_nw).dist_abs('GATTACA', 'GCATGCU'),
NeedlemanWunsch(1, _sim_nw).dist_abs('GATTACA', 'GCATGCU'),
)
self.assertEqual(
Gotoh(5, 5, _sim_wikipedia).dist_abs('AGACTAGTTAC', 'CGAGACGT'),
16)
self.assertGreaterEqual(
Gotoh(5, 2, _sim_wikipedia).dist_abs('AGACTAGTTAC', 'CGAGACGT'),
NeedlemanWunsch(5,
_sim_wikipedia).dist_abs('AGACTAGTTAC',
'CGAGACGT'),
)
# checked against http://ds9a.nl/nwunsch/ (mismatch=1, gap=5, skew=5)
self.assertEqual(
Gotoh(5, 5, _sim_nw).dist_abs('CGATATCAG', 'TGACGSTGC'), -5)
self.assertGreaterEqual(
Gotoh(5, 2, _sim_nw).dist_abs('CGATATCAG', 'TGACGSTGC'),
NeedlemanWunsch(5, _sim_nw).dist_abs('CGATATCAG', 'TGACGSTGC'),
)
self.assertEqual(
Gotoh(5, 5, _sim_nw).dist_abs('AGACTAGTTAC', 'TGACGSTGC'), -7)
self.assertGreaterEqual(
Gotoh(5, 2, _sim_nw).dist_abs('AGACTAGTTAC', 'TGACGSTGC'),
NeedlemanWunsch(5, _sim_nw).dist_abs('AGACTAGTTAC', 'TGACGSTGC'),
)
self.assertEqual(
Gotoh(5, 5, _sim_nw).dist_abs('AGACTAGTTAC', 'CGAGACGT'), -15)
self.assertGreaterEqual(
Gotoh(5, 2, _sim_nw).dist_abs('AGACTAGTTAC', 'CGAGACGT'),
NeedlemanWunsch(5, _sim_nw).dist_abs('AGACTAGTTAC', 'CGAGACGT'),
)
# Test wrapper
self.assertEqual(
gotoh('AGACTAGTTAC', 'CGAGACGT', 5, 5, _sim_wikipedia), 16)
def test_needleman_wunsch_sim(self):
"""Test abydos.distance.NeedlemanWunsch.sim."""
self.assertEqual(NeedlemanWunsch().sim('', ''), 1.0)
# https://en.wikipedia.org/wiki/Needleman–Wunsch_algorithm
self.assertEqual(
NeedlemanWunsch(1, _sim_nw).sim('GATTACA', 'GCATGCU'), 0)
self.assertEqual(
NeedlemanWunsch(5, _sim_wikipedia).sim('AGACTAGTTAC', 'CGAGACGT'),
0.19950186722152657,
)
# checked against http://ds9a.nl/nwunsch/ (mismatch=1, gap=5, skew=5)
nw5 = NeedlemanWunsch(5, _sim_nw)
self.assertEqual(nw5.sim('CGATATCAG', 'TGACGSTGC'), 0)
self.assertEqual(nw5.sim('AGACTAGTTAC', 'TGACGSTGC'), 0)
self.assertEqual(nw5.sim('AGACTAGTTAC', 'CGAGACGT'), 0)
class NeedlemanWunschTestCases(unittest.TestCase):
"""Test Needleman-Wunsch functions.
abydos.distance.NeedlemanWunsch
"""
cmp = NeedlemanWunsch()
def test_needleman_wunsch_dist_abs(self):
"""Test abydos.distance.NeedlemanWunsch.dist_abs."""
self.assertEqual(needleman_wunsch('', ''), 0)
# https://en.wikipedia.org/wiki/Needleman–Wunsch_algorithm
self.assertEqual(needleman_wunsch('GATTACA', 'GCATGCU', 1, _sim_nw), 0)
self.assertEqual(
needleman_wunsch('AGACTAGTTAC', 'CGAGACGT', 5, _sim_wikipedia), 16
)
# checked against http://ds9a.nl/nwunsch/ (mismatch=1, gap=5, skew=5)
self.assertEqual(
needleman_wunsch('CGATATCAG', 'TGACGSTGC', 5, _sim_nw), -5
)
self.assertEqual(
needleman_wunsch('AGACTAGTTAC', 'TGACGSTGC', 5, _sim_nw), -7
)
self.assertEqual(
needleman_wunsch('AGACTAGTTAC', 'CGAGACGT', 5, _sim_nw), -15
)
def test_needleman_wunsch_dist_abs_nialls(self):
"""Test abydos.distance.NeedlemanWunsch.dist_abs (Nialls set)."""
# checked against http://ds9a.nl/nwunsch/ (mismatch=1, gap=2, skew=2)
nw_vals = (5, 0, -2, 3, 1, 1, -2, -2, -1, -3, -3, -5, -3, -7, -7, -19)
for i in range(len(NIALL)):
self.assertEqual(
needleman_wunsch(NIALL[0], NIALL[i], 2, _sim_nw), nw_vals[i]
)
class GotohTestCases(unittest.TestCase):
"""Test Gotoh functions.
abydos.distance.Gotoh
"""
cmp = Gotoh()
nw = NeedlemanWunsch()
def test_gotoh_dist_abs(self):
"""Test abydos.distance.Gotoh.dist_abs."""
self.assertEqual(gotoh('', ''), 0)
# https://en.wikipedia.org/wiki/Needleman–Wunsch_algorithm
self.assertEqual(
self.cmp.dist_abs('GATTACA', 'GCATGCU', 1, 1, _sim_nw), 0)
self.assertGreaterEqual(
self.cmp.dist_abs('GATTACA', 'GCATGCU', 1, 0.5, _sim_nw),
self.nw.dist_abs('GATTACA', 'GCATGCU', 1, _sim_nw),
)
self.assertEqual(
self.cmp.dist_abs('AGACTAGTTAC', 'CGAGACGT', 5, 5, _sim_wikipedia),
16,
)
self.assertGreaterEqual(
self.cmp.dist_abs('AGACTAGTTAC', 'CGAGACGT', 5, 2, _sim_wikipedia),
self.nw.dist_abs('AGACTAGTTAC', 'CGAGACGT', 5, _sim_wikipedia),
)
# checked against http://ds9a.nl/nwunsch/ (mismatch=1, gap=5, skew=5)
self.assertEqual(
self.cmp.dist_abs('CGATATCAG', 'TGACGSTGC', 5, 5, _sim_nw), -5)
self.assertGreaterEqual(
self.cmp.dist_abs('CGATATCAG', 'TGACGSTGC', 5, 2, _sim_nw),
self.nw.dist_abs('CGATATCAG', 'TGACGSTGC', 5, _sim_nw),
)
self.assertEqual(gotoh('AGACTAGTTAC', 'TGACGSTGC', 5, 5, _sim_nw), -7)
self.assertGreaterEqual(
self.cmp.dist_abs('AGACTAGTTAC', 'TGACGSTGC', 5, 2, _sim_nw),
self.nw.dist_abs('AGACTAGTTAC', 'TGACGSTGC', 5, _sim_nw),
)
self.assertEqual(
self.cmp.dist_abs('AGACTAGTTAC', 'CGAGACGT', 5, 5, _sim_nw), -15)
self.assertGreaterEqual(
self.cmp.dist_abs('AGACTAGTTAC', 'CGAGACGT', 5, 2, _sim_nw),
self.nw.dist_abs('AGACTAGTTAC', 'CGAGACGT', 5, _sim_nw),
)
# Test wrapper
self.assertEqual(
gotoh('AGACTAGTTAC', 'CGAGACGT', 5, 5, _sim_wikipedia), 16)
def test_gotoh_dist_abs_nialls(self):
"""Test abydos.distance.Gotoh.dist_abs (Nialls set)."""
# checked against http://ds9a.nl/nwunsch/ (mismatch=1, gap=2, skew=2)
nw_vals = (5, 0, -2, 3, 1, 1, -2, -2, -1, -3, -3, -5, -3, -7, -7, -19)
for i in range(len(NIALL)):
self.assertEqual(
self.cmp.dist_abs(NIALL[0], NIALL[i], 2, 2, _sim_nw),
nw_vals[i],
)
nw_vals2 = (5, 0, -2, 3, 1, 1, -2, -2, -1, -2, -3, -3, -2, -6, -6, -8)
for i in range(len(NIALL)):
self.assertEqual(
self.cmp.dist_abs(NIALL[0], NIALL[i], 2, 1, _sim_nw),
nw_vals2[i],
)
self.assertGreaterEqual(
self.cmp.dist_abs(NIALL[0], NIALL[i], 2, 0.5, _sim_nw),
self.nw.dist_abs(NIALL[0], NIALL[i], 2, _sim_nw),
)