def test_combo__reverse_operations(self):
"""Tests a combination of insertions and deletions, similar to above,
but with operations applied in reverse order.
"""
before_deletion = 'AAAAA'
deleted_bases = 'TTT'
after_deletion = 'GGGGGGGGGGG'
insertion = 'AAA'
after_insertion = 'TTTTT'
raw_seq_str = (before_deletion + deleted_bases + after_deletion +
after_insertion)
seq = Seq(raw_seq_str, generic_dna)
seq_record = SeqRecord(seq)
maker = VCFToGenbankMaker(seq_record, None, None)
insertion_start = (len(before_deletion) + len(deleted_bases) +
len(after_deletion))
insertion_1_data = {'position': insertion_start, 'sequence': insertion}
maker.handle_insertion(insertion_1_data)
deletion_data = {
'interval':
(len(before_deletion), len(before_deletion) + len(deleted_bases))
}
maker.handle_deletion(deletion_data)
EXPECTED_SEQ = (before_deletion + after_deletion + insertion +
after_insertion)
self.assertEqual(len(EXPECTED_SEQ), len(seq_record.seq))
self.assertEqual(EXPECTED_SEQ, str(seq_record.seq))
def test_deletion__greater_than_one_base__3_bases(self):
"""Tests a deletion of 3 bases.
"""
before_deletion = 'AAAAA'
deleted_bases = 'TTT'
after_deletion = 'GGGGGGGGGGG'
raw_seq_str = before_deletion + deleted_bases + after_deletion
seq = Seq(raw_seq_str, generic_dna)
seq_record = SeqRecord(seq)
maker = VCFToGenbankMaker(seq_record, None, None)
deletion_data = {
'interval':
(len(before_deletion), len(before_deletion) + len(deleted_bases))
}
maker.handle_deletion(deletion_data)
# Assert the sequence is correct.
EXPECTED_SEQ = before_deletion + after_deletion
self.assertEqual(len(EXPECTED_SEQ), len(seq_record.seq))
self.assertEqual(EXPECTED_SEQ, str(seq_record.seq))
# Assert the liftover is correct.
EXPECTED_LIFTOVER = [((0, 4), (0, 4)), ((8, 18), (5, 15))]
self.assertEqual(EXPECTED_LIFTOVER,
maker.runtime_liftover._interval_mapping)
def test_deletion__multiple_positions(self):
"""Tests multiple deletions.
"""
before_deletion = 'AAAAA'
deleted_bases = 'TTT'
after_deletion = 'GGGGGGGGGGG'
other_deletion = 'C'
after_other_deletion = 'TTTTT'
raw_seq_str = (before_deletion + deleted_bases + after_deletion +
other_deletion + after_other_deletion)
seq = Seq(raw_seq_str, generic_dna)
seq_record = SeqRecord(seq)
maker = VCFToGenbankMaker(seq_record, None, None)
deletion_data = {
'interval':
(len(before_deletion), len(before_deletion) + len(deleted_bases))
}
maker.handle_deletion(deletion_data)
other_deletion_start = (len(before_deletion) + len(deleted_bases) +
len(after_deletion))
other_deletion_end = other_deletion_start + len(other_deletion)
other_deletion_data = {
'interval': (other_deletion_start, other_deletion_end)
}
maker.handle_deletion(other_deletion_data)
EXPECTED_SEQ = before_deletion + after_deletion + after_other_deletion
self.assertEqual(len(EXPECTED_SEQ), len(seq_record.seq))
self.assertEqual(EXPECTED_SEQ, str(seq_record.seq))
def test_combo(self):
"""Tests a combination of insertions, deletions, and snps.
"""
before_deletion = 'AAAAA'
deleted_bases = 'TTT'
after_deletion = 'GGGGGGGGGGG'
after_deletion_with_snp = 'GGGCGGGGGGG'
insertion = 'AAA'
after_insertion = 'TTTTT'
raw_seq_str = (before_deletion + deleted_bases + after_deletion +
after_insertion)
seq = Seq(raw_seq_str, generic_dna)
seq_record = SeqRecord(seq)
maker = VCFToGenbankMaker(seq_record, None, None)
# Make deletion.
deletion_data = {
'interval':
(len(before_deletion), len(before_deletion) + len(deleted_bases))
}
maker.handle_deletion(deletion_data)
# Make insertion.
insertion_start = (len(before_deletion) + len(deleted_bases) +
len(after_deletion))
insertion_1_data = {'position': insertion_start, 'sequence': insertion}
maker.handle_insertion(insertion_1_data)
# Make SNP.
insertion_start = (len(before_deletion) + len(deleted_bases) +
after_deletion_with_snp.index('C'))
snp_data = {'position': insertion_start, 'ref': 'G', 'alt': 'C'}
maker.handle_snp(snp_data)
EXPECTED_SEQ = (before_deletion + after_deletion_with_snp + insertion +
after_insertion)
self.assertEqual(len(EXPECTED_SEQ), len(seq_record.seq))
self.assertEqual(EXPECTED_SEQ, str(seq_record.seq))
def test_deletion__simple(self):
"""Tests a single deletion.
"""
before_deletion = 'AAAAA'
deleted_base = 'T'
after_deletion = 'GGGGGGGGGGG'
raw_seq_str = before_deletion + deleted_base + after_deletion
seq = Seq(raw_seq_str, generic_dna)
seq_record = SeqRecord(seq)
maker = VCFToGenbankMaker(seq_record, None, None)
deletion_data = {
'interval': (len(before_deletion), len(before_deletion) + 1)
}
maker.handle_deletion(deletion_data)
# Assert the sequence is correct.
EXPECTED_SEQ = before_deletion + after_deletion
self.assertEqual(EXPECTED_SEQ, str(seq_record.seq))
# Assert the liftover is correct.
EXPECTED_LIFTOVER = [((0, 4), (0, 4)), ((6, 16), (5, 15))]
self.assertEqual(EXPECTED_LIFTOVER,
maker.runtime_liftover._interval_mapping)