def kymogroups_2lines():
_, _, photon_count, parameters = read_dataset_gaussian(
"kymo_data_2lines.npz")
pixel_size = parameters[0].pixel_size
centers = [p.center / pixel_size for p in parameters]
image = CalibratedKymographChannel.from_array(photon_count,
pixel_size=pixel_size)
_, n_frames = image.data.shape
lines = KymoLineGroup([
KymoLine(np.arange(0.0, n_frames), np.full(n_frames, c), image)
for c in centers
])
# introduce gaps into tracked lines
use_frames = np.array([0, 1, -2, -1])
gapped_lines = KymoLineGroup([
KymoLine(line.time_idx[use_frames], line.coordinate_idx[use_frames],
image) for line in lines
])
# crop the ends of initial lines and make new set of lines with one cropped and the second full
truncated_lines = KymoLineGroup([
KymoLine(np.arange(1.0, n_frames - 2), np.full(n_frames - 3, c), image)
for c in centers
])
mixed_lines = KymoLineGroup([truncated_lines[0], lines[1]])
return lines, gapped_lines, mixed_lines
def test_stitching():
channel = CalibratedKymographChannel("test_data", np.array([[]]), 1e9, 1)
segment_1 = KymoLine([0, 1], [0, 1], channel)
segment_2 = KymoLine([2, 3], [2, 3], channel)
segment_3 = KymoLine([2, 3], [0, 0], channel)
segment_1b = KymoLine([0, 1], [0, 0], channel)
segment_1c = KymoLine([-1, 0, 1], [0, 0, 1], channel)
radius = 0.05
segment_1d = KymoLine([0.0, 1.0], [radius + 0.01, radius + 0.01], channel)
# Out of stitch range (maximum extension = 1)
assert len(
stitch_kymo_lines([segment_1, segment_3, segment_2], radius, 1,
2)) == 3
# Out of stitch radius
assert len(
stitch_kymo_lines([segment_1d, segment_3, segment_2], radius, 2,
2)) == 3
stitched = stitch_kymo_lines([segment_1, segment_3, segment_2], radius, 2,
2)
assert len(stitched) == 2
np.testing.assert_allclose(stitched[0].coordinate_idx, [0, 1, 2, 3])
np.testing.assert_allclose(stitched[1].coordinate_idx, [0, 0])
stitched = stitch_kymo_lines([segment_1b, segment_3, segment_2], radius, 2,
2)
np.testing.assert_allclose(stitched[0].coordinate_idx, [0, 0, 0, 0])
np.testing.assert_allclose(stitched[0].time_idx, [0, 1, 2, 3])
np.testing.assert_allclose(stitched[1].coordinate_idx, [2, 3])
# Check whether only the last two points are used (meaning we extrapolate [0, 0], [1, 1])
stitched = stitch_kymo_lines([segment_1c, segment_3, segment_2], radius, 2,
2)
np.testing.assert_allclose(stitched[0].coordinate_idx, [0, 0, 1, 2, 3])
np.testing.assert_allclose(stitched[0].time_idx, [-1, 0, 1, 2, 3])
# When using all three points, we shouldn't stitch
assert len(
stitch_kymo_lines([segment_1c, segment_3, segment_2], radius, 2,
3)) == 3
# Check whether the alignment has to work in both directions
# - and - should connect
l1, l2 = KymoLine([0, 1], [0, 0], channel), KymoLine([2, 2.01], [0, 0],
channel)
assert len(stitch_kymo_lines([l1, l2], radius, 1, 2)) == 1
# - and | should not connect.
l1, l2 = KymoLine([0, 1], [0, 0], channel), KymoLine([2, 2.01], [0, 1],
channel)
assert len(stitch_kymo_lines([l1, l2], radius, 1, 2)) == 2
def test_kymoline_interpolation():
time_idx = np.array([1.0, 3.0, 5.0])
coordinate_idx = np.array([1.0, 3.0, 3.0])
kymoline = KymoLine(time_idx, coordinate_idx, [])
interpolated = kymoline.interpolate()
np.testing.assert_allclose(interpolated.time_idx,
[1.0, 2.0, 3.0, 4.0, 5.0])
np.testing.assert_allclose(interpolated.coordinate_idx,
[1.0, 2.0, 3.0, 3.0, 3.0])
# Test whether concatenation still works after interpolation
np.testing.assert_allclose((interpolated + kymoline).time_idx,
[1.0, 2.0, 3.0, 4.0, 5.0, 1.0, 3.0, 5.0])
np.testing.assert_allclose((interpolated + kymoline).coordinate_idx,
[1.0, 2.0, 3.0, 3.0, 3.0, 1.0, 3.0, 3.0])
def test_refinement_2d():
time_idx = np.array([1, 2, 3, 4, 5])
coordinate_idx = np.array([1, 2, 3, 3, 3])
# Draw image with a deliberate offset
offset = 2
data = np.zeros((7, 7))
data[coordinate_idx + offset, time_idx] = 5
data[coordinate_idx - 1 + offset, time_idx] = 1
data[coordinate_idx + 1 + offset, time_idx] = 1
image = CalibratedKymographChannel.from_array(data)
line = KymoLine(time_idx[::2], coordinate_idx[::2], image=image)
refined_line = refine_lines_centroid([line], 5)[0]
np.testing.assert_allclose(refined_line.time_idx, time_idx)
np.testing.assert_allclose(refined_line.coordinate_idx,
coordinate_idx + offset)
# Test whether concatenation still works after refinement
np.testing.assert_allclose((refined_line + line).time_idx,
np.hstack((time_idx, time_idx[::2])))
np.testing.assert_allclose(
(refined_line + line).coordinate_idx,
np.hstack((coordinate_idx + offset, coordinate_idx[::2])),
)
def test_filter_lines():
channel = CalibratedKymographChannel("test_data", np.array([[]]), 1e9, 1)
k1 = KymoLine([1, 2, 3], [1, 2, 3], channel)
k2 = KymoLine([2, 3], [1, 2], channel)
k3 = KymoLine([2, 3, 4, 5], [1, 2, 4, 5], channel)
lines = KymoLineGroup([k1, k2, k3])
assert len(filter_lines(lines, 5)) == 0
assert all([
line1 == line2
for line1, line2 in zip(filter_lines(lines, 5), [k1, k3])
])
assert all([
line1 == line2
for line1, line2 in zip(filter_lines(lines, 2), [k1, k2, k3])
])
def test_refinement_line(loc, inv_sigma=0.3):
xx = np.arange(0, 50) - loc
image = np.exp(-inv_sigma * xx * xx)
calibrated_image = CalibratedKymographChannel.from_array(
np.expand_dims(image, 1))
line = refine_lines_centroid([KymoLine([0], [25], image=calibrated_image)],
5)[0]
np.testing.assert_allclose(line.coordinate_idx, loc, rtol=1e-2)
def test_stitch(kymograph, mockevent):
kymo_widget = KymoWidgetGreedy(kymograph, "red", 1, use_widgets=False)
k1 = KymoLine(
np.array([1, 2, 3]),
np.array([1, 1, 1]),
CalibratedKymographChannel.from_kymo(kymograph, "red"),
)
k2 = KymoLine(
np.array([6, 7, 8]),
np.array([3, 3, 3]),
CalibratedKymographChannel.from_kymo(kymograph, "red"),
)
kymo_widget.lines = KymoLineGroup([k1, k2])
# Go into line connection mode
kymo_widget._select_state({"new": "Connect Lines"})
assert len(kymo_widget.lines) == 2
in_um, in_s = calibrate_to_kymo(kymograph)
# Drag but stop too early (not leading to a connected line)
kymo_widget._line_connector.button_down(
mockevent(kymo_widget._axes, in_s(3), in_um(1), 3, 0))
kymo_widget._line_connector.button_release(
mockevent(kymo_widget._axes, in_s(4), in_um(3), 3, 0))
assert len(kymo_widget.lines) == 2
# Drag all the way (stitch the two)
kymo_widget._line_connector.button_down(
mockevent(kymo_widget._axes, in_s(3), in_um(1), 3, 0))
kymo_widget._line_connector.button_release(
mockevent(kymo_widget._axes, in_s(6), in_um(3), 3, 0))
# Verify the stitched line
np.testing.assert_allclose(kymo_widget.lines[0].time_idx,
[1, 2, 3, 6, 7, 8])
np.testing.assert_allclose(kymo_widget.lines[0].coordinate_idx,
[1, 1, 1, 3, 3, 3])
assert len(kymo_widget.lines) == 1
def test_kymoline_regression_sample_from_image_clamp():
"""This tests for a regression that occurred in sample_from_image. When sampling the image, we
sample pixels in a region around the line. This sampling procedure is constrained to stay within
the image. Previously, we used the incorrect axis to clamp the coordinate.
"""
# Sampling the bottom row of a three pixel tall image will return [0, 0] instead of [1, 3];
# since both coordinates would be clamped to the edge of the image (sampling nothing)."""
img = CalibratedKymographChannel("test_data",
np.array([[1, 1, 1], [3, 3, 3]]).T, 1e9,
1)
assert np.array_equal(
KymoLine([0, 1], [2, 2], img).sample_from_image(0), [1, 3])
def test_stitch_anywhere(start, stop, same_line, kymograph, mockevent):
kymo_widget = KymoWidgetGreedy(kymograph, "red", 1, use_widgets=False)
k1 = KymoLine(
np.array([1, 2, 3, 4, 5]),
np.array([1, 1, 1, 3, 3]),
CalibratedKymographChannel.from_kymo(kymograph, "red"),
)
k2 = KymoLine(
np.array([6, 7, 8]),
np.array([3, 3, 3]),
CalibratedKymographChannel.from_kymo(kymograph, "red"),
)
kymo_widget.lines = KymoLineGroup([k1, k2])
# Go into line connection mode
kymo_widget._select_state({"new": "Connect Lines"})
in_um, in_s = calibrate_to_kymo(kymograph)
# Merge points
kymo_widget._line_connector.button_down(
mockevent(kymo_widget._axes, in_s(start), in_um(1), 3, 0))
kymo_widget._line_connector.button_release(
mockevent(kymo_widget._axes, in_s(stop), in_um(3), 3, 0))
# Verify the stitched line
if not same_line:
time_result = np.hstack((np.arange(start) + 1, np.arange(stop, 9)))
coord_result = np.hstack((np.full(start, 1), np.full(9 - stop, 3)))
else:
time_result = np.hstack((np.arange(start) + 1, np.arange(stop, 6)))
coord_result = np.hstack((np.full(start, 1), np.full(6 - stop, 3)))
np.testing.assert_allclose(kymo_widget.lines[0].time_idx, time_result)
np.testing.assert_allclose(kymo_widget.lines[0].coordinate_idx,
coord_result)
assert len(kymo_widget.lines) == 2 if same_line else 1
def kymogroups_close_lines():
_, _, photon_count, parameters = read_dataset_gaussian(
"two_gaussians_1d.npz")
pixel_size = parameters[0].pixel_size
centers = [p.center / pixel_size for p in parameters]
image = CalibratedKymographChannel.from_array(photon_count,
pixel_size=pixel_size)
_, n_frames = image.data.shape
lines = KymoLineGroup([
KymoLine(np.arange(0.0, n_frames), np.full(n_frames, c), image)
for c in centers
])
return lines
def test_sampling():
test_data = np.array([
[0, 0, 0, 0, 0],
[0, 1, 0, 0, 0],
[0, 1, 1, 0, 0],
[0, 0, 1, 1, 0],
[0, 0, 1, 1, 0],
[0, 0, 0, 0, 0],
])
test_img = CalibratedKymographChannel("test", test_data, 10e9, 5)
# Tests the bound handling
kymoline = KymoLine([0, 1, 2, 3, 4], [0, 1, 2, 3, 4], test_img)
np.testing.assert_allclose(kymoline.sample_from_image(50), [0, 2, 3, 2, 0])
np.testing.assert_allclose(kymoline.sample_from_image(2), [0, 2, 3, 2, 0])
np.testing.assert_allclose(kymoline.sample_from_image(1), [0, 2, 2, 2, 0])
np.testing.assert_allclose(kymoline.sample_from_image(0), [0, 1, 1, 1, 0])
np.testing.assert_allclose(
KymoLine([0, 1, 2, 3, 4], [4, 4, 4, 4, 4],
test_img).sample_from_image(0), [0, 0, 1, 1, 0])
kymoline = KymoLine([0.1, 1.1, 2.1, 3.1, 4.1], [0.1, 1.1, 2.1, 3.1, 4.1],
test_img)
np.testing.assert_allclose(kymoline.sample_from_image(50), [0, 2, 3, 2, 0])
np.testing.assert_allclose(kymoline.sample_from_image(2), [0, 2, 3, 2, 0])
np.testing.assert_allclose(kymoline.sample_from_image(1), [0, 2, 2, 2, 0])
np.testing.assert_allclose(kymoline.sample_from_image(0), [0, 1, 1, 1, 0])
kymoline = KymoLine([0.1, 1.1, 2.1, 3.1, 4.1], [4.1, 4.1, 4.1, 4.1, 4.1],
test_img)
np.testing.assert_allclose(kymoline.sample_from_image(0), [0, 0, 1, 1, 0])