def test_sequences_blosum():
scores = read_substitution_matrix(
Path(__file__).parent / "rsrc" / "substitution.txt")
substitution = alignment.make_substitution_fn(scores)
s1 = "AGACTAGTTAC"
s2 = "CGAGACGT"
value, scores, paths = alignment.needleman_wunsch(
s1, s2, substitution=substitution)
algn, s1a1, s2a1 = alignment.best_alignment(paths, s1, s2, gap='-')
algn_sol1 = [list('--AGACTAGTTAC'), list('CGAGAC--G-T--')]
assert s1a1 == algn_sol1[0]
assert s2a1 == algn_sol1[1]
def test_substitution_function():
scores = read_substitution_matrix(
Path(__file__).parent / "rsrc" / "substitution.txt")
substitution = alignment.make_substitution_fn(scores, gap=0.5)
assert substitution('A', 'B') == (1, 0.5)
assert substitution('A', 'A') == (-20.0, 0.5)
assert substitution('A', 'C') == (3.0, 0.5)
# test just changing gap function
substitution2 = alignment.make_substitution_fn({}, gap=0.5)
assert substitution2('A', 'A') == (-1, 0.5)
assert substitution2('G', 'C') == (1, 0.5)
assert alignment._default_substitution_fn('A', 'A') == (-1, 1)
assert alignment._default_substitution_fn('G', 'C') == (1, 1)
def test_sequences_custom():
scores = read_substitution_matrix(
Path(__file__).parent / "rsrc" / "substitution.txt")
substitution = alignment.make_substitution_fn(scores)
s1 = "CCAGG"
s2 = "CCGA"
value, scores, paths = alignment.needleman_wunsch(s1, s2)
algn, s1a1, s2a1 = alignment.best_alignment(paths, s1, s2, gap='-')
algn_sol1 = [list("CCAGG"), list("CC-GA")]
assert value == 1
assert s1a1 == algn_sol1[0]
assert s2a1 == algn_sol1[1]
value, scores, paths = alignment.needleman_wunsch(
s1, s2, substitution=substitution)
algn, s1a2, s2a2 = alignment.best_alignment(paths, s1, s2, gap='-')
algn_sol2 = [list("CC-AGG"), list("CCGA--")]
assert s1a2 == algn_sol2[0]
assert s2a2 == algn_sol2[1]