def test_add_sequence(self):
contig = EMBLContig()
contig.header = self.create_blank_bit_of_contig()
contig.features['feature_1'] = self.create_blank_bit_of_contig()
contig.add_sequence('AAAACCCGGTNN')
self.assertIsInstance(contig.sequence, EMBLSequence)
self.assertRaises(ValueError, contig.add_sequence, 'AAAACCCGGTNN')
def visit_feature_node(self, feature_node):
sequence_id = feature_node.get_seqid()
contig = self.contigs.get(sequence_id)
if contig: # contig already exists, just try and update it
contig.add_feature(sequence_id=sequence_id,
feature_type=feature_node.get_type(),
start=feature_node.get_start(),
end=feature_node.get_end(),
strand=feature_node.get_strand(),
feature_attributes=feature_node.attribs,
locus_tag=self.locus_tag,
translation_table=self.translation_table)
else:
contig = EMBLContig()
successfully_added_feature = contig.add_feature(
sequence_id=sequence_id,
feature_type=feature_node.get_type(),
start=feature_node.get_start(),
end=feature_node.get_end(),
strand=feature_node.get_strand(),
feature_attributes=feature_node.attribs,
locus_tag=self.locus_tag,
translation_table=self.translation_table)
if successfully_added_feature:
self.contigs[sequence_id] = contig
else:
pass # discard the contig because we didn't add a feature so it is empty
def test_format_line_too_long(self):
contig = EMBLContig()
header_mock = MagicMock()
sequence_mock = MagicMock()
header_mock.format.return_value = "Header"
sequence_mock.format.return_value = "Very long line: " + '*'*100
contig.header = header_mock
contig.features = {}
contig.sequence = sequence_mock
self.assertRaises(ValueError, contig.format)
def test_add_feature(self):
contig = EMBLContig()
contig.add_feature(
sequence_id = 1,
feature_type = 'tRNA',
start = 100,
end = 200,
strand = '+',
feature_attributes = {'some_attribute': 'ABC' }
)
self.assertEqual(len(contig.features), 1)
def test_add_ignored_feature(self, feature_mock):
contig = EMBLContig()
feature_mock.return_value.format.side_effect = lambda: None
contig.add_feature(
sequence_id = 1,
feature_type = 'tRNA',
start = 100,
end = 200,
strand = '+',
feature_attributes = {'some_attribute': 'ABC' }
)
self.assertEquals(contig.features, {})
def visit_feature_node(self, feature_node):
sequence_id = feature_node.get_seqid()
contig = self.contigs.get(sequence_id)
if contig: # contig already exists, just try and update it
contig.add_feature(sequence_id = sequence_id, feature_type = feature_node.get_type(), start = feature_node.get_start(),
end = feature_node.get_end(), strand = feature_node.get_strand(),
feature_attributes = feature_node.attribs,
locus_tag = self.locus_tag, translation_table = self.translation_table)
else:
contig = EMBLContig()
successfully_added_feature = contig.add_feature(sequence_id = sequence_id, feature_type = feature_node.get_type(), start = feature_node.get_start(),
end = feature_node.get_end(), strand = feature_node.get_strand(),
feature_attributes = feature_node.attribs,
locus_tag = self.locus_tag, translation_table = self.translation_table)
if successfully_added_feature:
self.contigs[sequence_id] = contig
else:
pass # discard the contig because we didn't add a feature so it is empty
def test_format(self):
contig = EMBLContig()
header_mock = MagicMock()
feature_mock_1 = MagicMock()
feature_mock_2 = MagicMock()
sequence_mock = MagicMock()
header_mock.format.return_value = "Header\n"
feature_mock_1.format.return_value = "Feature 1\n"
feature_mock_1.start = 0
feature_mock_2.format.return_value = "Feature 2\n"
feature_mock_2.start = 100
sequence_mock.format.return_value = "Sequence\n"
contig.header = header_mock
contig.features = {1: feature_mock_1, 2: feature_mock_2}
contig.sequence = sequence_mock
calculated_string = contig.format()
expected_string = """\
Header
Feature 1
Feature 2
Sequence
"""
self.assertEqual(calculated_string, expected_string)
def test_add_header(self):
contig = EMBLContig()
contig.features['feature_1'] = self.create_blank_bit_of_contig()
contig.sequence = self.create_blank_bit_of_contig()
header_details = {
"authors": "John Doe",
"classification": "UNC",
"genome_type": "circular",
"organism": "My organism",
"project": "PRJ1234",
"publication": "Unpublished",
"sequence_identifier": "**contig123",
"sequence_length": 8,
"sequence_name": "chromX",
"taxon_id": 5678,
"title": "My title"
}
contig.add_header(**header_details)
self.assertIsInstance(contig.header, EMBLHeader)
with self.assertRaises(ValueError):
contig.add_header(**header_details)
def test_get_sorted_features_3(self):
# Must be able to get a list of features sorted by (start, end) irrespective of strand
contig = EMBLContig()
feature_1 = MagicMock()
feature_1.start = 100
feature_1.end = 1500
feature_2 = MagicMock()
feature_2.start = 1100
feature_2.end = 2000
contig.features = {1: feature_1, 2: feature_2}
expected_features = [feature_1, feature_2]
self.assertEquals(contig.sorted_features(), expected_features)
contig.features = {2: feature_2, 1: feature_1}
expected_features = [feature_1, feature_2]
self.assertEquals(contig.sorted_features(), expected_features)
contig.features = {2: feature_1, 1: feature_2}
expected_features = [feature_1, feature_2]
self.assertEquals(contig.sorted_features(), expected_features)
def visit_sequence_node(self, sequence_node):
sequence_id = sequence_node.get_description()
contig = self.contigs.setdefault(sequence_id, EMBLContig())
contig.add_sequence(sequence_node.get_sequence())
def test_format_no_header(self): contig = EMBLContig() contig.features['feature_1'] = self.create_blank_bit_of_contig() contig.sequence = self.create_blank_bit_of_contig() self.assertEqual(contig.header, None) self.assertRaises(ValueError, contig.format)
def test_format_no_features(self): contig = EMBLContig() contig.header = self.create_blank_bit_of_contig() contig.sequence = self.create_blank_bit_of_contig() self.assertEquals(contig.format(), '')
