def test_unaligned_muscle(self):
if not functions.which('muscle'):
_LOG.warning('muscle not found... skipping tests.')
return
seq1 = SeqRecord(Seq('AC--GTNAC-TYATR'), id='1')
seq2 = SeqRecord(Seq('ACN-GTAAC--CATT'), id='2')
# 'ACGTNACTYATR'
# 'ACNGTAACCATT'
diffs, l = seqstats.get_differences(seq1, seq2, aligned = False,
ignore_gaps = False,
aligner_tools = ['muscle'])
e = {11: ('R', 'T'),
3: ('T', 'G'),
6: ('C', 'A'),
7: ('T', 'C')}
self.assertEqual(diffs, e)
self.assertEqual(l, 12)
seq1 = SeqRecord(Seq('ATCCGT'), id='1')
seq2 = SeqRecord(Seq('ACCGT'), id='2')
diffs, l = seqstats.get_differences(seq1, seq2, aligned = False,
ignore_gaps = True,
aligner_tools = ['muscle'])
self.assertEqual(diffs, {})
self.assertEqual(l, 5)
diffs, l = seqstats.get_differences(seq1, seq2, aligned = False,
ignore_gaps = False,
aligner_tools = ['muscle'])
self.assertEqual(diffs, {1: ('T', '-')})
self.assertEqual(l, 6)
def test_unaligned_mafft_aa(self):
seq1 = SeqRecord(Seq('AC--DEFGI-LBATR'), id='1')
seq2 = SeqRecord(Seq('ACN-DEFGI--DATT'), id='2')
# 'AC-DEFGILBATR'
# 'ACNDEFGIDATT-'
e = {2: ('-', 'N'),
8: ('L', 'D'),
9: ('B', 'A'),
10: ('A', 'T'),
12: ('R', '-'),
}
diffs, l = seqstats.get_differences(seq1, seq2, aligned = False,
alphabet = alphabets.ProteinAlphabet(),
ignore_gaps = False,
aligner_tools = ['mafft'])
self.assertEqual(diffs, e)
self.assertEqual(l, 13)
diffs, l = seqstats.get_differences(seq1, seq2, aligned = False,
alphabet = alphabets.ProteinAlphabet(),
ignore_gaps = True,
aligner_tools = ['mafft'])
e = {8: ('L', 'D'),
9: ('B', 'A'),
10: ('A', 'T'),
}
self.assertEqual(diffs, e)
self.assertEqual(l, 11)
def test_align_error(self):
seq1 = SeqRecord(Seq('ACGT'), id='1')
seq2 = SeqRecord(Seq('TACGT'), id='2')
self.assertRaises(seqstats.AlignmentError, seqstats.get_differences,
seq1, seq2,
True)
diffs, l = seqstats.get_differences(seq1, seq2)
self.assertEqual(diffs, {})
self.assertEqual(l, 4)
expected = {0: ('-', 'T')}
diffs, l = seqstats.get_differences(seq1, seq2, ignore_gaps = False)
self.assertEqual(diffs, expected)
self.assertEqual(l, 5)
def test_aligned(self):
seq1 = SeqRecord(Seq('AC--GTNAC-TYATR'), id='1')
seq2 = SeqRecord(Seq('ACN-GTAAC--CATT'), id='2')
e = {14:('R', 'T')}
diffs, l = seqstats.get_differences(seq1, seq2, aligned = True)
self.assertEqual(diffs, e)
self.assertEqual(l, 11)
diffs, l = seqstats.get_differences(seq1, seq2, aligned = True,
ignore_gaps = False)
e = {14: ('R', 'T'),
2: ('-', 'N'),
10: ('T', '-')}
self.assertEqual(diffs, e)
self.assertEqual(l, 15)
def test_aligned_aa(self):
seq1 = SeqRecord(Seq('AC--DEFGI-LBATR'), id='1')
seq2 = SeqRecord(Seq('ACN-DEFGI--DATT'), id='2')
e = {14:('R', 'T')}
diffs, l = seqstats.get_differences(seq1, seq2, aligned = True,
alphabet = alphabets.ProteinAlphabet())
self.assertEqual(diffs, e)
self.assertEqual(l, 11)
diffs, l = seqstats.get_differences(seq1, seq2, aligned = True,
alphabet = alphabets.ProteinAlphabet(),
ignore_gaps = False)
e = {14: ('R', 'T'),
2: ('-', 'N'),
10: ('L', '-')}
self.assertEqual(diffs, e)
self.assertEqual(l, 15)
def test_unaligned(self):
seq1 = SeqRecord(Seq('AC--GTNAC-TYATR'), id='1')
seq2 = SeqRecord(Seq('ACN-GTAAC--CATT'), id='2')
diffs, l = seqstats.get_differences(seq1, seq2, aligned = False,
ignore_gaps = False,
aligner_tools = None)
e = {12: ('R', 'T'),
2: ('-', 'N'),
8: ('T', '-')}
self.assertEqual(diffs, e)
self.assertEqual(l, 13)
seq1 = SeqRecord(Seq('ATCCGT'), id='1')
seq2 = SeqRecord(Seq('ACCGT'), id='2')
diffs, l = seqstats.get_differences(seq1, seq2, aligned = False,
ignore_gaps = True,
aligner_tools = None)
self.assertEqual(diffs, {})
self.assertEqual(l, 5)
diffs, l = seqstats.get_differences(seq1, seq2, aligned = False,
ignore_gaps = False,
aligner_tools = None)
self.assertEqual(diffs, {1: ('T', '-')})
self.assertEqual(l, 6)