def test_find_overlap(self):
""" Test the find/overlap function with an easy case
"""
seq1 = "abdc sdf"
seq2 = "dfke "
seq3 = "dfs abd"
seq4 = "dc s"
seq5 = "pp"
position, n_overlaps = KMP.find_or_overlap(seq1, seq2)
self.assertEqual(n_overlaps, 2)
self.assertEqual(position, 6)
position, n_overlaps = KMP.find_or_overlap(seq3, seq1)
self.assertEqual(n_overlaps, 3)
self.assertEqual(position, 4)
position, n_overlaps = KMP.find_or_overlap(seq1, seq4)
self.assertEqual(n_overlaps, 4)
self.assertEqual(position, 2)
position, n_overlaps = KMP.find_or_overlap(seq1, seq5)
self.assertEqual(n_overlaps, 0)
self.assertEqual(position, len(seq1))
def compute_overlaps(fragments_dict, mat):
""" Compute the overlaps between the fragments and store them in a MatchManager
@param fragments_dict A dictionary of fragments with their ID as key
@param mat A MatchManager
"""
log.debug("Computing overlaps")
# sort the keys in descending order. A greater key will correspond (roughly) to
# a longer fragment. This condition is only guaranteed during the first iteration
keys = sorted(fragments_dict.keys())
keys.reverse()
for i, j in itertools.product(keys, keys):
if i == j:
continue
if not mat.needs_calculation(i, j, len(fragments_dict[i]), len(fragments_dict[j])):
continue
position, n_chars = KMP.find_or_overlap(fragments_dict[i], fragments_dict[j])
log.debug("Overlap between sequences %s (left) and %s (right): %s",i, j, n_chars)
m = Match(i,j, position, n_chars)
mat.store(m)
