def testGapAndMatch(self):
"""
Two sequences containing matches and gaps must compare as
expected.
"""
read1 = AARead('id1', 'GALHN-')
read2 = AARead('id2', 'GALHNA')
match = compareAaReads(read1, read2)
self.assertEqual(
'''\
Matches: 5/6 (83.33%)
Mismatches: 1/6 (16.67%)
Not involving gaps (i.e., conflicts): 0
Involving a gap in one sequence: 1/6 (16.67%)
Involving a gap in both sequences: 0
Id: id1
Length: 6
Gaps: 1/6 (16.67%)
Gap locations (1-based): 6
Id: id2
Length: 6
Gaps: 0''',
matchToString(match, read1, read2)
)
identicalLengths = len1 == len2
# Sanity check.
if args.align:
assert identicalLengths
match = compareAaReads(read1, read2, offsets=offsets)
x = 'Post-alignment, sequence' if args.align else 'Sequence'
if identicalLengths:
print('%s lengths are identical: %s' % (x, len1))
else:
print('%s lengths: %d, %d (difference %d)' % (x, len1, len2,
abs(len1 - len2)))
print(matchToString(match, read1, read2, offsets=offsets))
if args.showDiffs:
# Print all sites where the sequences differ.
width = int(log10(max(len1, len2))) + 1
headerPrinted = False
for site, (a, b) in enumerate(zip(read1.sequence, read2.sequence),
start=1):
if a != b:
if not headerPrinted:
print('Differences (site, %s, %s):' % (read1.id, read2.id))
headerPrinted = True
print(' %*d %s %s' % (width, site, a, b))
if not headerPrinted:
print('No sequence differences found.')
identicalLengths = len1 == len2
# Sanity check.
if args.align:
assert identicalLengths
match = compareAaReads(read1, read2, offsets=offsets)
x = 'Post-alignment, sequence' if args.align else 'Sequence'
if identicalLengths:
print('%s lengths are identical: %s' % (x, len1))
else:
print('%s lengths: %d, %d (difference %d)' %
(x, len1, len2, abs(len1 - len2)))
print(matchToString(match, read1, read2, offsets=offsets))
if args.showDiffs:
# Print all sites where the sequences differ.
width = int(log10(max(len1, len2))) + 1
headerPrinted = False
for site, (a, b) in enumerate(zip(read1.sequence, read2.sequence),
start=1):
if a != b:
if not headerPrinted:
print('Differences (site, %s, %s):' % (read1.id, read2.id))
headerPrinted = True
print(' %*d %s %s' % (width, site, a, b))
if not headerPrinted:
print('No sequence differences found.')