OMERO Zarr Microservice

An OMERO.server microservice that serves OME.zarr images and metadata based on the specification available at https://ngff.openmicroscopy.org/latest/. A list of implementations that can load data from omero-ms-zarr is available in the specification.

Summary

This microservice fetchs images from OMERO.server and converts them to OME.zarr images on the fly. It should be run alongside OMERO.server, and requires access to the OMERO PostgreSQL database and the OMERO data directory.

Clients can request an OMERO image over HTTP, and will receive an OME Zarr compliant image including metadata.

This microservice is still under heavy development, and does not yet support authentication. It is not suitable for production use.

Build

gradle build

if minio is not running, you can exclude the corresponding tests

gradle build -PexcludeTests="**/s3*"

Run

Copy your OMERO.server configuration etc/omero.properties to the microservice then,

gradle run --args=etc/omero.properties

and try

src/scripts/fetch-ms-zarr.py --endpoint-url http://localhost:8080/ --url-format '{url}image/{image}.zarr/' 1234

or

curl http://localhost:8080/image/1234.zarr/.zattrs

where 1234 is an image ID.

Configuration

In addition to your usual OMERO.server configuration, the microservice's etc/omero.properties may also include:

omero.ms.zarr.buffer-cache.size : pixel buffer cache size, default 16

omero.ms.zarr.chunk.size.adjust : ordered list of dimensions to adjust to increase chunk size, default XYZ; C and T are not offered

omero.ms.zarr.chunk.size.min : minimum chunk size (not guaranteed), default 1048576; applies before compression

omero.ms.zarr.compress.zlib.level : zlib compression level for chunks, default 6

omero.ms.zarr.folder.layout : for directory listings, default nested chunks, can be flattened; none disables directory listings

omero.ms.zarr.mask.split.enable : if masks split by ROI should be offered; default is false, can be set to true

omero.ms.zarr.mask.overlap.color : color to use for overlapping region in labeled masks, as a signed integer as in the OME Model; not set by default

omero.ms.zarr.mask.overlap.value : value to set for overlapping region in labeled masks, as a signed integer or "LOWEST" or "HIGHEST" (the default); setting to null disables overlap support

omero.ms.zarr.mask-cache.size : mask cache size in megabytes, default 250

omero.ms.zarr.net.path.image : URI template path for getting image data, default /image/{image}.zarr/ where {image} signifies the image ID and is mandatory

omero.ms.zarr.net.port : the TCP port on which the HTTP server should listen, default 8080

Build jar with dependencies

gradle shadowJar
java -jar build/libs/omero-ms-zarr-0.1.0-SNAPSHOT-all.jar etc/omero.properties

OMERO S3 Token Creator

To run as a standalone application you must define your access credentials using Java properties, environment variables, credentials files, or mechanisms specific to running on AWS infrastructure. See https://docs.aws.amazon.com/sdk-for-java/v2/developer-guide/credentials.html

For example, using Java properties, this will create a token for all objects (*) in bucket tmp :

java -Daws.accessKeyId=stsadmin -Daws.secretAccessKey=stsadmin-secret \
    -cp build/libs/omero-ms-zarr-0.1.0-SNAPSHOT-all.jar \
    org.openmicroscopy.s3.S3TokenCreator \
    -endpoint http://localhost:9000 -bucket tmp -prefix '*'

Example output:

{
  "endpoint_url": "http://localhost:9000",
  "region_name": "",
  "aws_access_key_id": "1234567890ABCDEFGHIJ",
  "aws_secret_access_key": "1234567890abcefghijklmnopqrdstuvxwzABCDE"
  "aws_session_token": "1234567890abcefghijklmnopqrdstuvxwzA.BCDEFGHIJKLMNOPQRSTUVWXYZ",
  "expiration": "2020-04-14T17:00:00Z"
}