Plates

[will-moore]

Test sets:

• https://minio-dev.openmicroscopy.org/idr/idr0033-rohban-pathways/41744_illum_corrected/pr_43/5966.zarr (big plate, 9 fields per well)
• https://minio-dev.openmicroscopy.org/idr/idr0002-heriche-condensation/plate1_1_013/pr_43/422.zarr (multi-T)
• https://minio-dev.openmicroscopy.org/idr/idr0004-thorpe-rad52/P101/pr_43/1751.zarr (sparse plate)
• https://minio-dev.openmicroscopy.org/idr/idr0001-graml-sysgro/JL_120731_S6A/pr_45/2551.zarr (multi-Z, multiple acquisitions)

To test:

• Install ome-zarr-py from this branch
• cd ome-zarr-py and pip install -e .
• Open plates (urls above) via e.g:

$ napari https://minio-dev.openmicroscopy.org/idr/idr0002-heriche-condensation/plate1_1_013/pr_43/422.zarr 

• Plates should be displayed in a grid with Well labels overlaid.
• You can view an individual Well by appending e.g. /A/1/ to the URL (actually, this will only work with the last plate above. The others were generated with an acquisition level in the hierarchy and need to add e.g. 0/A/1 or Run 422/A/1 for idr0002.

[will-moore]

Not doing very well with pre-commit here.
With pre-commit install I can't commit because flake8 is failing.
When I run flake8 locally it fails on a ton of files that I've not edited.

So I ran flake8 and fixed a bunch of errors, but when I try to commit them
the pre-commit hook fails and black undoes all the changes I just made!
So, reverted to pre-commit uninstall again!

[will-moore]

Exploring ways to add click handling to napari....
With ome-zarr-py using this branch.
Save this code to eg. plate.py:

import numpy as np
import napari
from string import ascii_uppercase


with napari.gui_qt():
    np.random.seed(1)
    viewer = napari.Viewer()

    # TODO: take this from command-line args
    # plate_url = "422.zarr".    # local usage
    plate_url = "https://s3.embassy.ebi.ac.uk/idr/share/community-call-2020-10-29/idr0002-heriche-condensation/plate1_1_013/422.zarr"

    field = "Field_1"

    plate_layers = viewer.open(plate_url)

    for plate_layer in plate_layers:

        @plate_layer.mouse_drag_callbacks.append
        def get_connected_component_shape(layer, event):
            # mouse-down event - ignore
            yield
            dragged = False
            # on move event
            while event.type == 'mouse_move':
                dragged = True
                yield
            # on release - ignore if user dragged
            if dragged:
                return

            click_x = layer.coordinates[-1]
            click_y = layer.coordinates[-2]

            # If user clicked on the Shapes layer, need to query other layers for Plate.shape
            # TODO: add the width and height to layer metadata
            for layer in viewer.layers:
                try:
                    plate_width = layer.data.shape[-1]
                    plate_height = layer.data.shape[-2]
                    break
                except Exception:
                    pass

            # Use 'plate' metadata to work out which Well was clicked
            print(layer.metadata)
            if "plate" in layer.metadata:
                plate = layer.metadata['plate']
                rows = plate['rows']
                columns = plate['columns']
                plateAcquisitions = plate.get('plateAcquisitions', [{'path': '0'}])
                run = plateAcquisitions[0]['path']
                print (rows, columns, plate_width, plate_height)
                well_width = plate_width // columns
                well_height = plate_height // rows
                row = int(click_y // well_height)
                column = int(click_x // well_width)
                print('row', row, 'column', column)

                # Open the Image from clicked Well
                field_url = f'{plate_url}/{run}/{ascii_uppercase[row]}/{column + 1}/{field}/'
                print('Open', field_url)
                well_name = f'{ascii_uppercase[row]}{column + 1}'
                napari.view_path(field_url)

Then run python plate.py

This loads the stitched plate.
When you click a Well (with one of the stitched plate layers selected), it works out which Well you clicked on and opens that Image - and renames the layers with the Well name. E.g. in this screenshot, A1 was clicked:

EDIT: updated to open Well in new window.

[will-moore]

That last commit adds Well labels and outlines (see screenshot above)

[joshmoore]

Nice!

[will-moore]

Running $ pre-commit run -a locally is passing now...

[will-moore]

Travis is failing because napari 0.4.0 release has dropped python 3.6

[will-moore]

With that last commit we can open a single Well, showing fields in a grid, using data exported with ome/omero-cli-zarr#43
e.g. 3 x 3 grid stitches together well in this example (but loading stitching together 9 x 1080 * 1080 images is very slow)

$ napari https://minio-dev.openmicroscopy.org/idr/idr0033-rohban-pathways/41744_illum_corrected/pr_43/5966.zarr/0/H/6/

[will-moore]

Showing a Plate exported with ome/omero-cli-zarr#43

napari https://minio-dev.openmicroscopy.org/idr/idr0094-ellinger-sarscov2/screenB/7825.zarr/

[joshmoore]

I generated a fake plate:

mkfake  plate -runs 2 -rows 2 -fields 2
omero import plate.fake/
omero zarr export Plate:101
napari 101.zarr/

and see:

👍 (No acquisitions expected yet, right?)

[will-moore]

I'm not yet filtering by acquisition here: e.g.
napari https://minio-dev.openmicroscopy.org/idr/idr0001-graml-sysgro/JL_120731_S6A/pr_45/2551.zarr/C/4/&acquisition=2661
filters by acquisition in vizarr but not in napari:

[joshmoore]

Thanks for getting rid of the print statements. Locally, napari -v 101.zarr # <- a plate! works fine. Getting this in.

[pwalczysko]

If I do

napari "https://minio-dev.openmicroscopy.org/idr/idr0002-heriche-condensation/plate1_1_013/pr_43/422.zarr/Run 422/A/1"

then it works and shows the well A1. But, when I play the timelapse, it seems to persistently play the frames between 8 and 18, and loop. Is there any explanation for that ? Update: When I clicked on the timebar, the movie jumped as expected to much higher framenumbers. When I clicked the timelapse play button again, it started to play the whole video, albeit in jumps of some 5 or 10 frames per displayed step, so no smoothness.

[pwalczysko]

on the sparse plate

napari https://minio-dev.openmicroscopy.org/idr/idr0004-thorpe-rad52/P101/pr_43/1751.zarr

18:30:21 ERROR Failed to load 0/D/12/0/3
18:30:21 ERROR Failed to load 0/G/11/0/3
18:30:21 ERROR Failed to load 0/H/10/0/3
18:30:21 ERROR Failed to load 0/F/12/0/3

Do we want to supress the message if and when there is no well ?