def test_03_02_get_measurement_columns_groups(self):
#
# Return the metadata grouping tags measurement if groups
#
groups = G.Groups()
groups.wants_groups.value = True
choices = ["Plate", "Well", "Site"]
for i, choice in enumerate(choices):
if i > 0:
groups.add_grouping_metadata()
del groups.grouping_metadata[i].metadata_choice.choices[:]
groups.grouping_metadata[i].metadata_choice.choices.extend(choices)
groups.grouping_metadata[i].metadata_choice.value = choice
columns = groups.get_measurement_columns(None)
self.assertEqual(len(columns), 4)
column = columns[0]
self.assertEqual(column[0], cpmeas.EXPERIMENT)
self.assertEqual(column[1], cpmeas.M_GROUPING_TAGS)
self.assertTrue(column[2].startswith(cpmeas.COLTYPE_VARCHAR))
column_metadata = []
for column in columns[1:]:
self.assertEqual(column[0], cpmeas.IMAGE)
self.assertEqual(column[2], cpmeas.COLTYPE_VARCHAR)
column_metadata.append(column[1])
for choice in choices:
self.assertTrue(cpmeas.C_METADATA + "_" +
choice in column_metadata)
def test_02_00_compute_no_groups(self):
groups, workspace = self.make_image_sets(
(
("Plate", ("P-12345", "P-23456")),
("Well", ("A01", "A02", "A03")),
("Site", ("1", "2", "3", "4")),
),
(
(
"DNA",
"Wavelength",
"1",
cpp.Pipeline.ImageSetChannelDescriptor.CT_GRAYSCALE,
),
(
"GFP",
"Wavelength",
"1",
cpp.Pipeline.ImageSetChannelDescriptor.CT_GRAYSCALE,
),
),
)
groups = G.Groups()
groups.wants_groups.value = False
m = workspace.measurements
assert isinstance(m, cpmeas.Measurements)
image_numbers = m.get_image_numbers()
expected_file_names = m[cpmeas.IMAGE, cpmeas.C_FILE_NAME + "_" + "DNA",
image_numbers]
self.assertTrue(groups.prepare_run(workspace))
self.assertEqual(len(image_numbers), 2 * 3 * 4)
output_file_names = m[cpmeas.IMAGE, cpmeas.C_FILE_NAME + "_" + "DNA",
image_numbers]
self.assertSequenceEqual(list(expected_file_names),
list(output_file_names))
def test_03_01_get_measurement_columns_nogroups(self):
#
# Don't return the metadata grouping tags measurement if no groups
#
groups = G.Groups()
groups.wants_groups.value = False
columns = groups.get_measurement_columns(None)
self.assertEqual(len(columns), 0)
def test_02_01_group_on_one(self):
groups = G.Groups()
groups, workspace = self.make_image_sets(
(("Plate", ("P-12345", "P-23456")),
("Well", ("A01", "A02", "A03")), ("Site", ("1", "2", "3", "4"))),
(("DNA", "Wavelength", "1",
cpp.Pipeline.ImageSetChannelDescriptor.CT_GRAYSCALE),
("GFP", "Wavelength", "1",
cpp.Pipeline.ImageSetChannelDescriptor.CT_GRAYSCALE)))
groups.wants_groups.value = True
groups.grouping_metadata[0].metadata_choice.value = "Plate"
groups.prepare_run(workspace)
m = workspace.measurements
assert isinstance(m, cpmeas.Measurements)
image_numbers = m.get_image_numbers()
self.assertEqual(len(image_numbers), 24)
np.testing.assert_array_equal(
np.hstack([np.ones(12, int),
np.ones(12, int) * 2]),
m[cpmeas.IMAGE, cpmeas.GROUP_NUMBER, image_numbers])
np.testing.assert_array_equal(
np.hstack([np.arange(1, 13)] * 2),
m[cpmeas.IMAGE, cpmeas.GROUP_INDEX, image_numbers])
pipeline = workspace.pipeline
assert isinstance(pipeline, cpp.Pipeline)
key_list, groupings = pipeline.get_groupings(workspace)
self.assertEqual(len(key_list), 1)
self.assertEqual(key_list[0], "Metadata_Plate")
self.assertEqual(len(groupings), 2)
for group_number, plate, (grouping, image_set_list) in zip(
(1, 2), ("P-12345", "P-23456"), groupings):
self.assertDictEqual(grouping, dict(Metadata_Plate=plate))
self.assertEqual(len(image_set_list), 3 * 4)
self.assertSequenceEqual(
list(image_set_list),
range((group_number - 1) * 12 + 1, group_number * 12 + 1))
for image_number in range(1 + (group_number - 1) * 12,
1 + group_number * 12):
for image_name in ("DNA", "GFP"):
ftr = "_".join((cpmeas.C_FILE_NAME, image_name))
self.assertTrue(m[cpmeas.IMAGE, ftr,
image_number].startswith(plate))
def make_image_sets(self, key_metadata, channel_metadata):
'''Make image sets by permuting key and channel metadata
key_metadata: collection of 2 tuples. 2 tuple is key + collection of
values, for instance, [("Well", ("A01", "A02", "A03")),
("Site", ("1","2","3","4"))]
channel_metadata: collection of tuple of channel name, channel metadata key,
channel metadata value and image type
returns a Groups module and a workspace with the image sets in its measurements
'''
iscds = [
cpp.Pipeline.ImageSetChannelDescriptor(channel_name, image_type)
for channel_name, metadata_key, metadata_value, image_type in
channel_metadata
]
image_set_key_names = [key for key, values in key_metadata]
slices = tuple([slice(0, len(values)) for key, values in key_metadata])
ijkl = np.mgrid[slices]
ijkl = ijkl.reshape(ijkl.shape[0], np.prod(ijkl.shape[1:]))
image_set_values = [values for key, values in key_metadata]
image_set_keys = [
tuple([
image_set_values[i][ijkl[i, j]] for i in range(ijkl.shape[0])
]) for j in range(ijkl.shape[1])
]
#
# The image set dictionary is a dictionary of image set key
# to a collection of ipds. We make up a file name that consists
# of the image set keys strung together with the channel metadata key.
# Likewise, the IPD metadata consists of the image set keys and
# the channel metadata key/value.
# Construct the dictionary by passing its constructor a collection
# of image_set_key / ipd collection tuples.
#
image_sets = []
for k in image_set_keys:
image_set = []
image_sets.append(image_set)
for _, channel_metadata_key, channel_metadata_value, _ in channel_metadata:
file_name = "_".join(list(k) +
[channel_metadata_value]) + ".tif"
metadata = dict([(kn, kv) for kn, kv in zip(
image_set_key_names + [channel_metadata_key],
list(k) + [channel_metadata_value])])
image_set.append((file_name, metadata))
#
# Scramble the image sets
#
r = np.random.RandomState()
r.seed(
np.frombuffer("".join(["%s=%s" % kv for kv in key_metadata]),
np.uint8))
image_sets = [image_sets[i] for i in r.permutation(len(image_sets))]
m = cpmeas.Measurements()
m.set_metadata_tags(
["_".join((cpmeas.C_METADATA, k)) for k in image_set_key_names])
m.set_channel_descriptors(iscds)
for i, ipds in enumerate(image_sets):
image_number = i + 1
for (file_name, metadata), iscd in zip(ipds, iscds):
for feature, value in ((cpmeas.C_FILE_NAME,
file_name), (cpmeas.C_PATH_NAME,
os.pathsep + "images"),
(cpmeas.C_URL,
"file://images/" + file_name)):
m[cpmeas.IMAGE, feature + "_" + iscd.name,
image_number] = value
for key, value in metadata.iteritems():
feature = "_".join((cpmeas.C_METADATA, key))
m[cpmeas.IMAGE, feature, image_number] = value
pipeline = cpp.Pipeline()
module = G.Groups()
module.module_num = 1
pipeline.add_module(module)
workspace = cpw.Workspace(pipeline, module, m, None, m, None)
return module, workspace
