def test_insert_sequence_no_features(self):
seq = Seq('ATGTTTGGG', generic_dna)
seq_record = SeqRecord(seq)
insert_seq = Seq('CCC', generic_dna)
insert_start_position = 6
updated_seq_record = insert_sequence(seq_record, insert_seq,
insert_start_position)
self.assertEqual('ATGTTTCCCGGG', str(updated_seq_record.seq))
def test_insert_sequence_no_features(self):
seq = Seq('ATGTTTGGG', generic_dna)
seq_record = SeqRecord(seq)
insert_seq = Seq('CCC', generic_dna)
insert_start_position = 6
updated_seq_record = insert_sequence(
seq_record, insert_seq, insert_start_position)
self.assertEqual('ATGTTTCCCGGG', str(updated_seq_record.seq))
def test_insert_sequence_update_features(self):
seq = Seq('ATGTTTGGGAAATTT', generic_dna)
seq_record = SeqRecord(seq)
feature_id_to_seq_map = {
1: 'ATG',
2: 'GGG',
3: 'AAATTT'
}
def _assert_feature_seq(feature, seq_record):
original_seq = feature_id_to_seq_map[feature.id]
self.assertEqual(original_seq, str(feature.extract(seq_record).seq))
# Make a mix of features, some before, some after the insertion.
feature_1_loc = FeatureLocation(0, 3)
feature_1 = SeqFeature(feature_1_loc, type='CDS', id=1)
seq_record.features.append(feature_1)
_assert_feature_seq(feature_1, seq_record)
feature_2_loc = FeatureLocation(6, 9)
feature_2 = SeqFeature(feature_2_loc, type='CDS', id=2)
seq_record.features.append(feature_2)
_assert_feature_seq(feature_2, seq_record)
feature_3_loc = FeatureLocation(9, 15)
feature_3 = SeqFeature(feature_3_loc, type='CDS', id=3)
seq_record.features.append(feature_3)
_assert_feature_seq(feature_3, seq_record)
# The sequence to insert.
insert_seq = Seq('CCC', generic_dna)
insert_start_position = 6
# Perform the insertion.
updated_seq_record = insert_sequence(
seq_record, insert_seq, insert_start_position)
# Assert conditions that we want to preserve.
self.assertEqual('ATGTTTCCCGGGAAATTT', str(updated_seq_record.seq))
self.assertEqual(
len(seq_record.features), len(updated_seq_record.features))
for feature in updated_seq_record.features:
_assert_feature_seq(feature, updated_seq_record)
def test_insert_sequence_update_features(self):
seq = Seq('ATGTTTGGGAAATTT', generic_dna)
seq_record = SeqRecord(seq)
feature_id_to_seq_map = {1: 'ATG', 2: 'GGG', 3: 'AAATTT'}
def _assert_feature_seq(feature, seq_record):
original_seq = feature_id_to_seq_map[feature.id]
self.assertEqual(original_seq,
str(feature.extract(seq_record).seq))
# Make a mix of features, some before, some after the insertion.
feature_1_loc = FeatureLocation(0, 3)
feature_1 = SeqFeature(feature_1_loc, type='CDS', id=1)
seq_record.features.append(feature_1)
_assert_feature_seq(feature_1, seq_record)
feature_2_loc = FeatureLocation(6, 9)
feature_2 = SeqFeature(feature_2_loc, type='CDS', id=2)
seq_record.features.append(feature_2)
_assert_feature_seq(feature_2, seq_record)
feature_3_loc = FeatureLocation(9, 15)
feature_3 = SeqFeature(feature_3_loc, type='CDS', id=3)
seq_record.features.append(feature_3)
_assert_feature_seq(feature_3, seq_record)
# The sequence to insert.
insert_seq = Seq('CCC', generic_dna)
insert_start_position = 6
# Perform the insertion.
updated_seq_record = insert_sequence(seq_record, insert_seq,
insert_start_position)
# Assert conditions that we want to preserve.
self.assertEqual('ATGTTTCCCGGGAAATTT', str(updated_seq_record.seq))
self.assertEqual(len(seq_record.features),
len(updated_seq_record.features))
for feature in updated_seq_record.features:
_assert_feature_seq(feature, updated_seq_record)
def insert_frt_site(genome_record,
upstream_frt_seq,
upstream_insert_pos,
downstream_frt_seq,
downstream_insert_pos,
feature_id_prefix='',
upstream_validation_seq=None,
downstream_validation_seq=None):
"""Inserts FRT sites at the given position.
Args:
genome_record: The genome_record to start with. This is not mutated
in this method.
upstream_frt: One of FRT_OPTIONS.
upstream_insert_pos: Insert position for the upstream FRT site.
downstream_frt: One of FRT_OPTIONS.
downstream_insert_pos: Insert position for the downstream FRT site.
NOTE: This is the insert position before the upstream is inserted.
This method will account for that.
feature_id_prefix: Prefix for the BioPython feature id.
Recommended: 'seg_3' for the upstream FRT site of
segment 3.
upstream_validation_seq: The next n bases after the start of the
upstream FRT site. This is used as a sanity check to make sure the
FRT is being inserted in the right place.
downstream_validation_seq: The n bases before the downstream site.
Used for sanity checking the insertion.
Returns:
A modified SeqRecord with the change.
"""
# Maybe check the validation bases.
if upstream_validation_seq:
actual_upstream_seq = str(
genome_record.seq[upstream_insert_pos:upstream_insert_pos +
len(upstream_validation_seq)])
assert upstream_validation_seq == actual_upstream_seq, (
"Actual %s" % actual_upstream_seq)
if downstream_validation_seq:
actual_downstream_seq = str(
genome_record.
seq[downstream_insert_pos -
len(downstream_validation_seq):downstream_insert_pos])
assert downstream_validation_seq == actual_downstream_seq, (
"Actual %s" % actual_downstream_seq)
# Adjust the downstream insert position to account for the upstream
# one being inserted first.
adjusted_downstream_insert_pos = (downstream_insert_pos +
len(upstream_frt_seq))
# Make a copy of the record so we can modify it.
updated_genome_record = copy.deepcopy(genome_record)
# Insert the upstream frt.
updated_genome_record = insert_sequence(
updated_genome_record,
upstream_frt_seq,
upstream_insert_pos,
safe_features=[],
insert_feature_type=FRT_FEATURE_TYPE,
insert_feature_id=feature_id_prefix + '_upstream_frt',
insert_feature_strand=1)
# Insert the downstream frt.
updated_genome_record = insert_sequence(
updated_genome_record,
downstream_frt_seq,
adjusted_downstream_insert_pos,
safe_features=[],
insert_feature_type=FRT_FEATURE_TYPE,
insert_feature_id=feature_id_prefix + '_downstream_frt',
insert_feature_strand=1)
return updated_genome_record