Note

Go to the end to download the full example code.

Coordinate Transform#

This script writes two positions using the high content screening (HCS) OME-Zarr dataset with two FOV in a single well with different coordinate transformations (i.e translation and scaling)

import os
from tempfile import TemporaryDirectory

import numpy as np

from iohub.ngff import TransformationMeta, open_ome_zarr

Set storage path

tmp_dir = TemporaryDirectory()
store_path = os.path.join(tmp_dir.name, "transformed.zarr")
print("Zarr store path", store_path)

Zarr store path /tmp/tmpkcacpph0/transformed.zarr

Create two random sample images

tczyx_1 = np.random.randint(
    0, np.iinfo(np.uint16).max, size=(1, 3, 3, 32, 32), dtype=np.uint16
)
tczyx_2 = np.random.randint(
    0, np.iinfo(np.uint16).max, size=(1, 3, 3, 32, 32), dtype=np.uint16
)

Coordinate Transformations (T,C,Z,Y,X) By default the translation is the identity matrix

coords_shift = [[1.0, 1.0, 1.0, 10.0, 10.0], [1.0, 1.0, 0.0, -10.0, -10.0]]
img_scaling = [[1.0, 1.0, 1.0, 0.5, 0.5]]

Generate Transformation Metadata

translation = []
for shift in coords_shift:
    translation.append(
        TransformationMeta(type="translation", translation=shift)
    )
scaling = []
for scale in img_scaling:
    scaling.append(TransformationMeta(type="scale", scale=scale))

Write 5D data to a new Zarr store

with open_ome_zarr(
    store_path,
    layout="hcs",
    mode="w-",
    channel_names=["DAPI", "GFP", "Brightfield"],
) as dataset:
    # Create and write to positions
    # This affects the tile arrangement in visualization
    position = dataset.create_position(0, 0, 0)
    position.create_image("0", tczyx_1, transform=[translation[0]])
    position = dataset.create_position(0, 0, 1)
    position.create_image("0", tczyx_2, transform=[translation[1], scaling[0]])
    # Print dataset summary
    dataset.print_tree()

/
 └── 0
     └── 0
         ├── 0
         │   └── 0 (1, 3, 3, 32, 32) uint16
         └── 1
             └── 0 (1, 3, 3, 32, 32) uint16

Note

To see the coordinate transforms, open the positions individually using napari-ome-zarr. This will duplicate the layers (channels).

Clean up

tmp_dir.cleanup()
Show Source