def test_alignment(self):
self.assertEqual(align_glocal("test", 'test'), (0, 0))
self.assertEqual(align_glocal("test", ' test'), (0, 1))
self.assertEqual(align_glocal(" test", 'test'), (1, 0))
self.assertEqual(align_glocal("tast", 'test'), (1, 0))
self.assertEqual(align_glocal("teste", 'tast'), (2, 0))
self.assertEqual(align_glocal("test", ' tast'), (1, 1))
def find_ranges(needle, genome, max_edit):
"""
Finds all matches in the FM-index genome for the needle within max_edit edit distance
"""
#Calculate the biggest substring that must exist in the haystack if the needle matches
# with the edit distance. This assumes worst case distribution of deletion-edits.
k = (len(needle) - max_edit) / (max_edit + 1)
#Generate a generator of all substrings of length k.
kmers = (i for i in xrange(len(needle) - k + 1))
#Skip all the hard work if the edit distance is zero.
if not max_edit:
return [(h, 0) for h in genome.search(needle)], 0
#Find where all of these kmers match exactly
kmerhits = (genome.search(needle[kmer:kmer + k]) for kmer in kmers)
#Create all a sorted set of intervals
ranges = SortedSet(updator=OverlappingIntervalsUpdator)
#Iterate over all of the kmers and kmers matches
for i, kmer in enumerate(kmerhits):
for hit in kmer:
#Check for any existing possible ranges that are already in our list
overlaps = ranges.overlap_point(hit)
if not len(overlaps):
#Create a new range consisting of the worst-case given the position of the kmer in the needle
ranges.add([hit - i - max_edit, hit - i + len(needle) + max_edit, 1])
else:
#Incrememnt the number of kmers in each possible range
for overlap in overlaps:
overlap[2] += 1
best_edit = max_edit
#Iterate over all potential alignments and use dynamic programming to determine whether it actually
#fits within edit distance
#Possible matches defined as the ranges with > max_edit distance kmers found in the match
alignments = []
for potential_alignment in (r for r in ranges if r[2] > max_edit):
ed, start = align_glocal(needle, genome.seq[potential_alignment[0]:potential_alignment[1]])
if ed < best_edit:
best_edit = ed
alignments = []
alignments.append((potential_alignment[0] + start, ed))
return alignments, best_edit
