def test_infer_align_format(self):
# simple cases
# map
line = 'S1/1 NC_123456'
obs = infer_align_format(StringIO(line))
self.assertEqual(obs[0], 'map')
self.assertListEqual(obs[1], [line])
# b6o
line = 'S1/1 NC_123456 100 100 0 0 1 100 25 124 0.1 100'
self.assertEqual(infer_align_format(StringIO(line))[0], 'b6o')
# sam
line = 'S1 77 NC_123456 26 0 100M * 0 0 * *'
self.assertEqual(infer_align_format(StringIO(line))[0], 'sam')
# sam header
line = '@HD VN:1.0 SO:unsorted'
self.assertEqual(infer_align_format(StringIO(line))[0], 'sam')
# empty file
with self.assertRaises(ValueError) as ctx:
infer_align_format(StringIO())
self.assertEqual(str(ctx.exception), (
'Alignment file is empty or unreadable.'))
# invalid sam
line = 'S1 * * * * * * 0 0 * *'
with self.assertRaises(ValueError) as ctx:
infer_align_format(StringIO(line))
self.assertEqual(str(ctx.exception), (
'Cannot determine alignment file format.'))
# cannot determine
line = 'Hi there!'
with self.assertRaises(ValueError) as ctx:
infer_align_format(StringIO(line))
self.assertEqual(str(ctx.exception), (
'Cannot determine alignment file format.'))
# real files
# Bowtie2 (sam)
with openzip(join(self.datdir, 'align', 'bowtie2', 'S01.sam.xz')) as f:
self.assertEqual(infer_align_format(f)[0], 'sam')
# BURST (b6o)
with openzip(join(self.datdir, 'align', 'burst', 'S01.b6.bz2')) as f:
self.assertEqual(infer_align_format(f)[0], 'b6o')
def test_openzip(self):
text = 'Hello World!'
# read regular file
fp = join(self.tmpdir, 'test.txt')
with open(fp, 'w') as f:
f.write(text)
with openzip(fp) as f:
obs = f.read()
self.assertEqual(obs, text)
# write regular file
with openzip(fp, 'wt') as f:
f.write('Here I am!')
with open(fp, 'r') as f:
obs = f.read()
self.assertEqual(obs, 'Here I am!')
remove(fp)
# read compressed file
fpz = join(self.tmpdir, 'test.txt.gz')
with gzip.open(fpz, 'wb') as f:
f.write(text.encode())
with openzip(fpz) as f:
obs = f.read()
self.assertEqual(obs, text)
remove(fpz)
# write compressed file
fpb = join(self.tmpdir, 'test.txt.bz2')
with openzip(fpb, 'at') as f:
f.write('Here I am!')
with bz2.open(fpb, 'rt') as f:
obs = f.read()
self.assertEqual(obs, 'Here I am!')
remove(fpb)
def test_read_gene_coords(self):
# simple case
tbl = ('## GCF_000123456', '# NC_123456', '1 5 384', '2 410 933',
'# NC_789012', '1 912 638', '2 529 75')
obs = read_gene_coords(tbl, sort=True)
exp = {
'NC_123456': [(5, True, True, '1'), (384, False, True, '1'),
(410, True, True, '2'), (933, False, True, '2')],
'NC_789012': [(75, True, True, '2'), (529, False, True, '2'),
(638, True, True, '1'), (912, False, True, '1')]
}
self.assertDictEqual(obs, exp)
# don't sort
obs = read_gene_coords(tbl, sort=False)['NC_789012']
exp = [(638, True, True, '1'), (912, False, True, '1'),
(75, True, True, '2'), (529, False, True, '2')]
self.assertListEqual(obs, exp)
# incorrect formats
# only one column
msg = 'Cannot extract coordinates from line:'
with self.assertRaises(ValueError) as ctx:
read_gene_coords(('hello', ))
self.assertEqual(str(ctx.exception), f'{msg} "hello".')
# only two columns
with self.assertRaises(ValueError) as ctx:
read_gene_coords(('hello\t100', ))
self.assertEqual(str(ctx.exception), f'{msg} "hello\t100".')
# three columns but 3rd is string
with self.assertRaises(ValueError) as ctx:
read_gene_coords(('hello\t100\tthere', ))
self.assertEqual(str(ctx.exception), f'{msg} "hello\t100\tthere".')
# real coords file
fp = join(self.datdir, 'function', 'coords.txt.xz')
with openzip(fp) as f:
obs = read_gene_coords(f, sort=True)
self.assertEqual(len(obs), 107)
obs_ = obs['G000006745']
self.assertEqual(len(obs_), 7188)
self.assertTupleEqual(obs_[0], (372, True, True, '1'))
self.assertTupleEqual(obs_[1], (806, False, True, '1'))
self.assertTupleEqual(obs_[2], (816, True, True, '2'))
self.assertTupleEqual(obs_[3], (2177, False, True, '2'))
