View OME-ZARR plates in Napari

This tutorial demonstrates how to load data generated with Fractal into Napari using operetta-compose and the ome-zarr-navigator

Note

If you use the napari-feature-classifier make sure to install napari=0.4.18 as higher version introduced a change in the colormap update mechanism which currently slows down classification.

conda create -n drp-env python=3.11
conda activate drp-env
pip install "napari[all]"==0.4.18 napari-feature-classifier napari-ome-zarr-navigator napari-feature-visualization

import operetta_compose as oc
import napari
import napari_ome_zarr_navigator
from pathlib import Path

def add_well(zarr_root: str, row: str, col: str, viewer: napari.viewer.Viewer):
    """Add a well with labels to the napari viewer

    Args:
        zarr_root: the base URL of the OME-ZARR
        row: the row of the plate to load
        col: the column of the plate to load
        viewer: an instance of a napari Viewer
    """
    zarr_url = str(Path(f"{zarr_root}/{row}/{col}/0").resolve())
    roi_url, roi_idx = oc.io.get_roi(zarr_url, "well_ROI_table", level=0)
    img = oc.io.load_intensity_roi(roi_url, roi_idx)
    labels = oc.io.load_label_roi(roi_url, roi_idx)

    viewer.add_image(img, name=f"{row}{col}")
    label_layer = viewer.add_labels(labels, name=f"{row}{col}_label")
    napari_ome_zarr_navigator.util.add_features_to_labels(zarr_url, label_layer)

Start a Napari Viewer and load a predefined well from the OME-Zarr

viewer = napari.Viewer()

row = "C"
col = "3"
zarr_root = "../../operetta-compose/tests/test_output/operetta_plate.zarr"

add_well(zarr_root, row, col, viewer)

Utilities

Inspect the classifier

import pickle
with open("../../operetta-compose/tests/classifier.pkl", "rb") as clf_file:
    classifier = pickle.load(clf_file)

Inspect predictions in the regionprops table

import anndata as ad
ad.read_zarr(f"{zarr_root}/{row}/{col}/0/tables/regionprops").obs

	roi_id	label	prediction
0	C3	1	2
1	C3	2	2
2	C3	3	1
3	C3	4	2
4	C3	5	1
...	...	...	...
3630	C3	3631	1
3631	C3	3632	2
3632	C3	3633	2
3633	C3	3634	1
3634	C3	3635	1

3635 rows × 3 columns