ome-zarr-io

Read, write, and validate OME-Zarr 0.5 images and high-content-screening plates: from a NumPy array to a spec-compliant OME-Zarr image in a few lines of Python.

Microscopy datasets are growing faster than desktop tools can handle. OME-Zarr is the community format for storing them in a chunked, cloud-friendly way that viewers such as napari and Vizarr can open. Getting the metadata exactly right is fiddly. ome-zarr-io handles it for you, so you can spend your time on the analysis instead of the file format.

ome-zarr-io writes NumPy or Dask arrays as OME-Zarr 0.5 images with axes, units, pixel sizes, multiscale pyramids, channels and segmentation labels, and arranges such images into high-content-screening plates. It also reads images back and validates images and plates against the OME-Zarr 0.5 schemas, from Python or the command line.

Installation

pip install ome-zarr-io

Quick start

import numpy as np
from ome_zarr_io import Writer, Reader, validate

image = np.random.randint(0, 255, size=(2, 16, 256, 256), dtype=np.uint8)

Writer(
    path="example.ome.zarr", image=image, dims=["c", "z", "y", "x"],
    axis_units={"z": "micrometer", "y": "micrometer", "x": "micrometer"},
    scale_transformations={"z": 0.5, "y": 0.2, "x": 0.2},
    channels={"DAPI": {"color": "0000FF"}, "GFP": {"color": "00FF00"}},
    downscale_levels=2, overwrite=True,
).write()

print(validate("example.ome.zarr"))                      # spec check
print(Reader("example.ome.zarr").get_voxel_size())       # physical pixel size

The same validation is available on the command line:

ome-zarr-io validate example.ome.zarr

Documentation

The full documentation is at https://turku-bioimaging.github.io/ome-zarr-io/:

• Getting started: installation, including remote (URL) support, and the terminology used throughout.
• Guides for reading and writing images, channels, labels, high-content-screening plates, validation and the command line.
• The API reference.