def test_show_examples_raises_on_wrong_column_labels(self):
output_prefix = test_utils.test_tmpfile('column_labels')
FLAGS.examples = testdata.GOLDEN_TRAINING_EXAMPLES
FLAGS.output = output_prefix
FLAGS.column_labels = 'read base,base quality,mapping quality,strand'
with self.assertRaisesRegex(ValueError, '--column_labels'):
show_examples.run()
# With 6 channel names, it should run without error:
FLAGS.column_labels = '1,2,3,4,5,6'
show_examples.run()
def test_show_examples_end2end_training_examples(self):
output_prefix = test_utils.test_tmpfile('training')
FLAGS.examples = testdata.GOLDEN_TRAINING_EXAMPLES
FLAGS.output = output_prefix
show_examples.run()
ls = glob.glob('{}*'.format(output_prefix))
filenames = [os.path.basename(path) for path in ls]
self.assertTrue(
all(['training_channels' in filename for filename in filenames]))
self.assertTrue(
all([filename.endswith('.png') for filename in filenames]))
self.assertTrue(all(['label' in filename for filename in filenames]),
msg='Training examples should produce labeled images.')
def test_show_examples_end2end_all_optional_parameters(self):
# Set all the optional parameters to check that they all work together.
output_prefix = test_utils.test_tmpfile('kitchen_sink')
FLAGS.examples = testdata.GOLDEN_TRAINING_EXAMPLES
FLAGS.output = output_prefix
FLAGS.annotation = False
FLAGS.regions = 'chr20:10,003,650-10,005,000'
FLAGS.vcf = testdata.TRUTH_VARIANTS_VCF
FLAGS.image_type = 'both'
FLAGS.num_records = 5
FLAGS.verbose = False
FLAGS.truth_labels = False # On by default for training examples.
show_examples.run()
ls = glob.glob('{}*'.format(output_prefix))
filenames = [os.path.basename(path) for path in ls]
self.assertTrue(
any([
'kitchen_sink_channels' in filename for filename in filenames
]),
msg='image_type=both, so there should be images with "channels"')
self.assertTrue(
any(['kitchen_sink_rgb' in filename for filename in filenames]),
msg='image_type=both, so there should be images with "rgb"')
self.assertTrue(
all([filename.endswith('.png') for filename in filenames]))
self.assertFalse(any(['label' in filename for filename in filenames]),
msg='Should be no "label" when truth_labels=False')
self.assertLen(filenames,
10,
msg='Should be 10 filenames, i.e. 5 records with '
'channels+rgb for each')
self.assertCountEqual(
filenames, [
'kitchen_sink_rgb_chr20:10004146_A->G.png',
'kitchen_sink_channels_chr20:10004146_A->G.png',
'kitchen_sink_rgb_chr20:10004093_A->C.png',
'kitchen_sink_channels_chr20:10004093_A->C.png',
'kitchen_sink_rgb_chr20:10003831_G->A.png',
'kitchen_sink_channels_chr20:10003831_G->A.png',
'kitchen_sink_rgb_chr20:10003691_A->G.png',
'kitchen_sink_channels_chr20:10003691_A->G.png',
'kitchen_sink_rgb_chr20:10003650_T->C.png',
'kitchen_sink_channels_chr20:10003650_T->C.png'
],
msg=
'Specific examples and their output filenames should be the same '
'if the inputs are the same.')