def test_GivenLongSequencesWithDiffAboveThreshold_AreNotRefLike(self):
sequences = ["AATTA", "AATTT", "TATTA", "GCGGG"]
seqlen = len(sequences[0])
self.assertTrue(seqlen >= LENGTH_THRESHOLD)
self.assertEqual(get_one_ref_like_threshold_distance(seqlen), 1)
self.assertEqual(get_majority_string(sequences), "AATTA")
self.assertFalse(sequences_are_one_reference_like(sequences))
def test_GivenLongSequencesWithDiffBelowThreshold_AreRefLike(self):
sequences = ["AATTA", "AATTT", "TATTA"]
# Sequences are above length threshold
seqlen = len(sequences[0])
self.assertTrue(seqlen >= LENGTH_THRESHOLD)
# We tolerate up to one diff against majority string
self.assertEqual(get_one_ref_like_threshold_distance(seqlen), 1)
# Here's the majority string
self.assertEqual(get_majority_string(sequences), "AATTA")
# No sequence is > distance threshold from the majority string
self.assertTrue(sequences_are_one_reference_like(sequences))
def test_majority_string_correct(self):
actual = get_majority_string(self.sequences)
expected = "GGGGCGGGCCC"
self.assertEqual(actual, expected)
def test_get_all_distances_to_majority_string(self):
majority_string = get_majority_string(self.sequences)
actual = list(get_distances(self.sequences, majority_string))
expected = [11, 11, 0, 1, 1, 0]
self.assertEqual(actual, expected)
def test_majority_string_bad_lengths_fails(self):
sequences = ["AAA-", "AT-", "AAAA"]
with self.assertRaises(ValueError):
get_majority_string(sequences)