def test_close_sortsAndIndexes(self):
with TempDirectory() as tmp_dir:
bam_path = os.path.join(tmp_dir.path, 'destination.bam')
header = { 'HD': {'VN': '1.0'},
'SQ': [{'LN': 1575, 'SN': 'chr1'},
{'LN': 1584, 'SN': 'chr2'}] }
align1 = ConnorAlign(mock_align(query_name='align1',
reference_start=100))
align2 = ConnorAlign(mock_align(query_name='align2',
reference_start=200))
align3 = ConnorAlign(mock_align(query_name='align3',
reference_start=300))
tag1 = BamTag('X1','Z', 'desc',
get_value=lambda family, pair, align: family)
tag2 = BamTag('X2','Z', 'desc',
get_value=lambda family, pair, align: align.query_name)
writer = samtools.AlignWriter(header, bam_path, [tag1, tag2])
writer.write('familyC', None, align3)
writer.write('familyA', None, align1)
writer.write('familyB', None, align2)
writer.close()
bamfile = samtools.alignment_file(bam_path, 'rb')
actual_aligns = [a for a in bamfile.fetch()]
bamfile.close()
self.assertEqual(3, len(actual_aligns))
self.assertEqual('align1', actual_aligns[0].query_name)
self.assertEqual('align2', actual_aligns[1].query_name)
self.assertEqual('align3', actual_aligns[2].query_name)
def test_write_removesTagsWhenValueIsNone(self):
with TempDirectory() as tmp_dir:
bam_path = os.path.join(tmp_dir.path, 'destination.bam')
header = { 'HD': {'VN': '1.0'},
'SQ': [{'LN': 1575, 'SN': 'chr1'},
{'LN': 1584, 'SN': 'chr2'}] }
align1 = ConnorAlign(mock_align(query_name='align1'))
align1.set_tag('X1', 'No', 'Z')
tag1 = BamTag('X1','Z', 'desc',
get_value = lambda family, pair, align: None)
writer = samtools.AlignWriter(header, bam_path, [tag1])
writer.write('familyA', None, align1)
writer.close()
bamfile = samtools.alignment_file(bam_path, 'rb')
actual_aligns = [a for a in bamfile.fetch()]
bamfile.close()
align_tags = {}
for actual_align in actual_aligns:
for t_name, t_val, t_type in actual_align.get_tags(with_value_type=True):
key = (actual_align.query_name, t_name)
t_type = AlignWriterTest.fix_pysam_inconsistent_tag_type(t_type)
align_tags[key] = "{}:{}:{}".format(t_name, t_type, t_val)
self.assertEqual(1, len(actual_aligns))
self.assertEqual(0, len(align_tags))
def test_init_defaultToNoTags(self):
with TempDirectory() as tmp_dir:
bam_path = os.path.join(tmp_dir.path, "destination.bam")
header = { 'HD': {'VN': '1.0'},
'SQ': [{'LN': 1575, 'SN': 'chr1'},
{'LN': 1584, 'SN': 'chr2'}] }
writer = samtools.AlignWriter(header, bam_path)
writer.close()
self.assertEqual([], writer._tags)
def test_close_logs(self):
with TempDirectory() as tmp_dir:
bam_path = os.path.join(tmp_dir.path, 'destination.bam')
header = { 'HD': {'VN': '1.0'},
'SQ': [{'LN': 1575, 'SN': 'chr1'},
{'LN': 1584, 'SN': 'chr2'}] }
align1 = ConnorAlign(mock_align(query_name='align1',
reference_start=100))
writer = samtools.AlignWriter(header, bam_path, [])
writer.write('familyA', None, align1)
writer.close(log=self.mock_logger)
info_log_lines = self.mock_logger._log_calls['INFO']
self.assertEqual(1, len(info_log_lines))
self.assertRegexpMatches(info_log_lines[0], 'destination.bam')
def test_write_addsAlignTags(self):
with TempDirectory() as tmp_dir:
bam_path = os.path.join(tmp_dir.path, 'destination.bam')
header = { 'HD': {'VN': '1.0'},
'SQ': [{'LN': 1575, 'SN': 'chr1'},
{'LN': 1584, 'SN': 'chr2'}] }
align1 = ConnorAlign(mock_align(query_name='align1'))
align2 = ConnorAlign(mock_align(query_name='align2'))
align3 = ConnorAlign(mock_align(query_name='align3'))
tag1 = BamTag('X1','Z', 'desc',
get_value=lambda family,pair,align: family)
tag2 = BamTag('X2','Z', 'desc',
get_value=lambda family,pair,align: pair)
tag3 = BamTag('X3','Z', 'desc',
get_value=lambda family,pair,align: align.query_name)
writer = samtools.AlignWriter(header, bam_path, [tag1, tag2, tag3])
writer.write('familyA', 'pair1', align1)
writer.write('familyB', 'pair2', align2)
writer.write('familyC', 'pair3', align3)
writer.close()
bamfile = samtools.alignment_file(bam_path, 'rb')
actual_aligns = [a for a in bamfile.fetch()]
bamfile.close()
align_tags = {}
for actual_align in actual_aligns:
for t_name, t_val, t_type in actual_align.get_tags(with_value_type=True):
key = (actual_align.query_name, t_name)
t_type = AlignWriterTest.fix_pysam_inconsistent_tag_type(t_type)
align_tags[key] = "{}:{}:{}".format(t_name, t_type, t_val)
self.assertEqual(3, len(actual_aligns))
self.assertEqual("X1:Z:familyA", align_tags[('align1', 'X1')])
self.assertEqual("X1:Z:familyB", align_tags[('align2', 'X1')])
self.assertEqual("X1:Z:familyC", align_tags[('align3', 'X1')])
self.assertEqual("X2:Z:pair1", align_tags[('align1', 'X2')])
self.assertEqual("X2:Z:pair2", align_tags[('align2', 'X2')])
self.assertEqual("X2:Z:pair3", align_tags[('align3', 'X2')])
self.assertEqual("X3:Z:align1", align_tags[('align1', 'X3')])
self.assertEqual("X3:Z:align2", align_tags[('align2', 'X3')])
self.assertEqual("X3:Z:align3", align_tags[('align3', 'X3')])
def test_write(self):
with TempDirectory() as tmp_dir:
bam_path = os.path.join(tmp_dir.path, "destination.bam")
header = { 'HD': {'VN': '1.0'},
'SQ': [{'LN': 1575, 'SN': 'chr1'},
{'LN': 1584, 'SN': 'chr2'}] }
align1 = ConnorAlign(mock_align(query_name="align1"))
align2 = ConnorAlign(mock_align(query_name="align2"))
align3 = ConnorAlign(mock_align(query_name="align3"))
family = None
writer = samtools.AlignWriter(header, bam_path)
writer.write(family, None, align1)
writer.write(family, None, align2)
writer.write(family, None, align3)
writer.close()
bamfile = samtools.alignment_file(bam_path, 'rb')
actual_query_names = [align.query_name for align in bamfile.fetch()]
bamfile.close()
self.assertEqual(['align1', 'align2', 'align3'], actual_query_names)
def test_write_addsHeaderTags(self):
with TempDirectory() as tmp_dir:
bam_path = os.path.join(tmp_dir.path, 'destination.bam')
header = { 'HD': {'VN': '1.0'},
'SQ': [{'LN': 1575, 'SN': 'chr1'},
{'LN': 1584, 'SN': 'chr2'}],
'CO': ['comment1', 'comment2']}
tag1 = BamTag('X1','Z', 'annotates family', get_value=None)
tag2 = BamTag('X2','Z', 'annotates alignment', get_value=None)
writer = samtools.AlignWriter(header, bam_path, [tag2, tag1])
writer.close()
bamfile = samtools.alignment_file(bam_path, 'rb')
actual_header = dict(bamfile.header)
bamfile.close()
expected_header = deepcopy(header)
expected_header.pop('CO')
actual_comments = actual_header.pop('CO')
expected_comments = ['comment1',
'comment2',
'connor\tBAM tag\tX1: annotates family',
'connor\tBAM tag\tX2: annotates alignment']
self.assertEqual(expected_comments, actual_comments)