def test_diff_length_seq(self):
config_file_mock = StringIO(
'{"GP": -2, "SAME": 2, "DIFF": -3, "MAX_SEQ_LENGTH": 10, "MAX_PATHS": 5}'
)
config = Needelman_Wunch.Config(config_file_mock)
seq1 = 'ADB'
seq2 = 'AB'
table = Needelman_Wunch.NwTable(seq1, seq2, config)
paths_generator = table.path_generator()
score, paths = next(paths_generator)
self.assertEqual(score, 2)
self.assertSequenceEqual(paths, ('ADB', 'A_B'))
with self.assertRaises(StopIteration):
next(paths_generator)
def test_output_limit(self):
config_file_mock = StringIO(
'{"GP": -2, "SAME": 2, "DIFF": -5, "MAX_SEQ_LENGTH": 10, "MAX_PATHS": 1}'
)
expected_seqs = {('AB_', 'A_D'), ('A_B', 'AD_')}
config = Needelman_Wunch.Config(config_file_mock)
seq1 = 'AB'
seq2 = 'AD'
table = Needelman_Wunch.NwTable(seq1, seq2, config)
paths_generator = table.path_generator()
score1, out_seqs1 = next(paths_generator)
self.assertEqual(score1, -2)
self.assertTrue({out_seqs1}.issubset(expected_seqs))
with self.assertRaises(StopIteration):
next(paths_generator)
def test_long_sequences(self):
# get all output paths
# sample comes from internet
config_file_mock = StringIO(
'{"GP": -2, "SAME": 1, "DIFF": -1, "MAX_SEQ_LENGTH": 15, "MAX_PATHS": 1000}'
)
expected_seq = {('GAA__CGA_ACGTAACC', 'CAAGACGACACG_AAC_')}
config = Needelman_Wunch.Config(config_file_mock)
seq1 = 'GAACGAACGTAACC'
seq2 = 'CAAGACGACACGAAC'
pats_set = set()
table = Needelman_Wunch.NwTable(seq1, seq2, config)
for score, paths in table.path_generator():
self.assertEqual(score, 0)
pats_set.add(paths)
self.assertTrue(expected_seq.issubset(pats_set))