In [1]:

from pathlib import Path
import utils
import trackmate_utils
import holoviews as hv
import panel as pn
pn.extension()

from pathlib import Path import utils import trackmate_utils import holoviews as hv import panel as pn pn.extension()

1. Load & View Data¶

Define data paths, read stacks & view a single frame

In [2]:

base_data_path = Path(r"data/fucci_60_frames")
red_stack_path = base_data_path / 'red.tif'
green_stack_path = base_data_path / 'green.tif'

red_stack = utils.read_stack(red_stack_path)
green_stack = utils.read_stack(green_stack_path)

base_data_path = Path(r"data/fucci_60_frames") red_stack_path = base_data_path / 'red.tif' green_stack_path = base_data_path / 'green.tif' red_stack = utils.read_stack(red_stack_path) green_stack = utils.read_stack(green_stack_path)

In [3]:

utils.view_stacks([red_stack, green_stack],
                  frame=0)

utils.view_stacks([red_stack, green_stack], frame=0)

Out[3]:

2. Segmentation (CellPose)¶

Define paths to CellPose models and segment each channel individually

In [4]:

base_model_path = Path(r"models")
red_segmentation_model = base_model_path / 'cellpose/nuclei_red_v2'
green_segmentation_model = base_model_path / 'cellpose/nuclei_green_v2'

base_model_path = Path(r"models") red_segmentation_model = base_model_path / 'cellpose/nuclei_red_v2' green_segmentation_model = base_model_path / 'cellpose/nuclei_green_v2'

In [5]:

# red_segmented = utils.segment_stack(red_stack_path, red_segmentation_model)
# green_segmented = utils.segment_stack(green_stack_path, green_segmentation_model)

# red_segmented = utils.segment_stack(red_stack_path, red_segmentation_model) # green_segmented = utils.segment_stack(green_stack_path, green_segmentation_model)

Read & view the segmentation map, where each color indicates a distinct cell

In [6]:

red_segmented_stack = utils.read_stack(base_data_path / 'red_segmented.tiff')
green_segmented_stack = utils.read_stack(base_data_path / 'green_segmented.tiff')
utils.view_segmented_stacks([red_segmented_stack, green_segmented_stack],
                            frame=0)

red_segmented_stack = utils.read_stack(base_data_path / 'red_segmented.tiff') green_segmented_stack = utils.read_stack(base_data_path / 'green_segmented.tiff') utils.view_segmented_stacks([red_segmented_stack, green_segmented_stack], frame=0)

Out[6]:

Overlay segmentation map over raw data

In [7]:

utils.view_segmentation_overlay([red_stack, green_stack],
                                [red_segmented_stack, green_segmented_stack],
                                frame=1)

utils.view_segmentation_overlay([red_stack, green_stack], [red_segmented_stack, green_segmented_stack], frame=1)

Out[7]:

3. Individual Cell Tracking (TrackMate)¶

Define paths to segmentation maps, TrackMate config file and run tracking

In [8]:

trackmate_settings_xml = base_model_path / Path(r"trackmate/basic_settings.xml")
red_segmented_path = base_data_path / 'red_segmented.tiff'
green_segmented_path = base_data_path / 'green_segmented.tiff'

# utils.run_trackmate(trackmate_settings_xml, red_segmented_path)
# utils.run_trackmate(trackmate_settings_xml, green_segmented_path)

trackmate_settings_xml = base_model_path / Path(r"trackmate/basic_settings.xml") red_segmented_path = base_data_path / 'red_segmented.tiff' green_segmented_path = base_data_path / 'green_segmented.tiff' # utils.run_trackmate(trackmate_settings_xml, red_segmented_path) # utils.run_trackmate(trackmate_settings_xml, green_segmented_path)

Define paths to tracking results and load them into a Pandas DataFrames

In [9]:

tm_red = trackmate_utils.TrackmateXML(base_data_path / "red_segmented.tiff.xml")
tm_green = trackmate_utils.TrackmateXML(base_data_path / "green_segmented.tiff.xml")

tm_red = trackmate_utils.TrackmateXML(base_data_path / "red_segmented.tiff.xml") tm_green = trackmate_utils.TrackmateXML(base_data_path / "green_segmented.tiff.xml")

TrackMate produces two datasets:

• Spots table - includes a unique identifier for each cell and some measurements (X, Y position of center of mass, perimeter, etc). Furthermore, it includes the frame a given cell appeared in and the grayscale value it got assigned during segmentation. Finally, an ROI is provided in the form of a polygon.
• Tracks table - includes a unique identifier for each track and the unique cell identifier of cells belonging to the track

As an example, we can take a look at these tables below:

In [10]:

tm_red.spots.head()

tm_red.spots.head()

Out[10]:

	frame	POSITION_X	POSITION_Y	PERIMETER	image_id	AREA	ROI
ID
1280	0	298.457614	98.531397	68.87987300699037	14	318.5	[(2.0423861852433447, 11.968602825745677), (6....
1281	0	249.164397	109.666342	94.16607520078527	15	514.0000000000003	[(11.33560311284046, 11.833657587548643), (13....
1282	0	51.860286	116.851993	106.62086195196227	16	723.5	[(-8.360285648468093, 12.648007371573357), (-6...
1283	0	8.230815	124.985312	98.34159973787956	17	556.0000000000001	[(-2.7308153477218227, 20.514688249400493), (-...
1284	0	306.240462	123.778112	70.29699476237073	18	331.99999999999994	[(1.259538152610446, 11.72188755020079), (3.25...

In [11]:

tm_red.tracks.head()

tm_red.tracks.head()

Out[11]:

	EDGE_TIME	TrackID	SPOT_SOURCE_ID	SPOT_TARGET_ID	EDGE_X_LOCATION	EDGE_Y_LOCATION
0	6.5	0	1441	252	334.9808900546602	28.096977698755694
1	5.5	0	67	1441	329.9621126644713	38.75997067240739
2	7.5	0	252	74	342.69301090695916	21.153634470310546
3	3.5	0	669	27	325.2599715099715	64.72873722138428
4	0.5	0	1280	106	303.10427771717116	93.1131562142458

Assuming we care about all the cells belonging to a specific track, a convenience method exists in the form of:

In [12]:

tm_red.trace_track(track_id=0)

tm_red.trace_track(track_id=0)

Out[12]:

	ID	frame	POSITION_X	POSITION_Y	PERIMETER	image_id	AREA	ROI
0	1280	0	298.457614	98.531397	68.87987300699037	14	318.5	[(2.0423861852433447, 11.968602825745677), (6....
1	106	1	307.750942	87.694915	80.22959146711749	11	354.0	[(4.74905838041434, 12.805084745762713), (5.74...
2	268	2	316.042723	75.078873	80.52926568031265	10	355.0000000000001	[(2.4572769953051647, 14.421126760563382), (3....
3	669	3	323.914279	70.212921	74.4187648052274	9	331.5	[(1.5857214680744391, 14.287078934137767), (3....
4	27	4	326.605664	59.244553	70.75207550444621	12	306.0	[(6.894335511982604, 12.255446623093682), (7.8...
5	67	5	328.896209	45.199491	81.61131907656338	8	360.5	[(5.603791030975515, 14.300508552935739), (6.6...
6	1441	6	331.028016	32.32045	86.08285487170572	6	407.49999999999994	[(9.471983640081817, 13.179550102249486), (10....
7	252	7	338.933764	23.873505	80.01652756195931	5	348.5	[(4.566236250597797, 13.626494500239122), (7.5...
8	74	8	346.452258	18.433763	82.65887799040118	6	387.50000000000017	[(7.047741935483884, 12.066236559139785), (11....
9	1315	9	354.105107	9.54788	74.48954469058977	4	365.49999999999994	[(5.394892840857267, 10.952120383036936), (6.3...
10	1377	10	356.991179	4.355153	51.972803760081376	5	179.50000000000006	[(3.508820798514421, 7.144846796657381), (4.50...

Which returns a subset table representing just the cells belonging to a particular track.

4. View Tracks¶

TODO restructure track views¶

In [13]:

%load_ext autoreload
%autoreload 2

%load_ext autoreload %autoreload 2

In [60]:

import trackmate_utils

trackmate_utils.view_track([red_stack, green_stack], frame=0, track=tm_red.trace_track(track_id=0))

import trackmate_utils trackmate_utils.view_track([red_stack, green_stack], frame=0, track=tm_red.trace_track(track_id=0))

Out[60]:

In [64]:

import trackmate_utils

trackmate_utils.view_side_by_side([red_stack, green_stack], frame=0, track=tm_red.trace_track(track_id=0))

import trackmate_utils trackmate_utils.view_side_by_side([red_stack, green_stack], frame=0, track=tm_red.trace_track(track_id=0))

Out[64]:

In [ ]: