2020-08-04

Attending: Jason, Josh, Petr, Wilma, J-m (Notes), Mark, Frances, June, Andreas, Dominik,

Talley Lambert, Simon, Kevin, Erin, Chris, Juna Nuñez-Iglesias, Melissa

Start: 2:00 pm

1. Accepting minutes from last meeting

2. Project Status

(2-3 minutes each)

• IDR

  • Preparing for next release:

    • Another covid dataset depending on time + 1 or 2 studies

      depending on testing

• NGFF

  • Masks status

    • [Final

      PR</u>](https://github.com/ome/omero-ms-zarr/pull/60) opened for the first round of labeled images incl. Color support, etc.

      • On disk named as “masks”, name to be modified

    • No support for overlap or multiscales

    • In investigating multiscales, unsure what viewers are going

      to do

  • Masks future work

    • Likely to change name/folder from “masks” to “labels”

    • Add in multiscale

    • Extend colors specification

    • Support links to external data sources

  • Metadata (provenance, etc.) - repeat of [last

    week</u>](https://docs.google.com/document/d/1HLpIUVJQrfH_km9D9ibqF2pFCrXS5yCS-bC9LjShpvQ/edit)

    • Location for storing “mask author” (ORCID, etc)

      • Provenance: who did what, using which tool/version

    • In terms of getting the JSON formatting correct, will also

      look at basic metadata:

      • Image name, physical pixel size

      • Dimension order, channel names,

      • Copyright information (IDR)

  • Napari feedback (Juan & Talley)

    • T: imaging what workflow pipelines might look like (in

      napari, etc.). Don’t have an internal metadata model. There’s just a dict where stuff can be dropped. No knowledge of dimensions. What happens if 2 separate nodes want to know something about a label? OME model would be a natural thing to standardize on. (Dimensionality)

    • J: hard-coding.

      • Example data to try out?

    • Josh: ask at regular intervals where the images are.

    • Juan: hierarchical model:

      • Josh: going in the direction of more hierarchies

      • Chris: balancing issue on filesystem

      • Mattias working on Zarr spec (czi grant): not sure if

        the current state could fix issues. Spec meeting on Friday

    • J-m: re: pipelines – running headless? Talley: always

      headless first.

      • Juan: strong desire is that “napari headless is python”
    • J-m: re: pipelines – any algorithms and what metadata they

      need?

      • Juan: all of scikit-image. They should be annotated with

        type information

      • Potentially something like an OME:Image type (but start

        simple)

      • I.e. know as much as possible about the image as

        possible but still work from raw arrays.

      • Talley: progressive enhancement. Start with dask plan,

        type & shape of array, and then identify algorithms.

      • J-m: how to determine the array for a filter in

        scikit-image.

      • Juan: should also be pure python (no GUI)

      • Chris: opencv guys have had image types for some time.

        I.e. some lite wrapping of arrays. Juan: type annotations rather than classes.

• Deep Copy (Josh &

  Mark)

  • Hybrid version which uses in-place like functionality to link

    (or if necessary copy) files in /Files as well as /ManagedRepository.

  • No database upgrade required.

  • Goal is to have working functionality before Mark leaves mid

    October.

  • Simon: anything under managedRepo is immutable? If so it should

    be fine to hardlink.

    • Hardlinking attachments which can be modified could be a

      problem (one file but two copies of metadata)

  • J-M: projection:

    • Duplicate of metadata might help to solve the metadata issue

  • Simon: copy vs symlink

    • Josh: decide where in the hierarchy where we can copy or

      symlink.

  • Petr: importance of documentation in terms of explanation

  • Deep copy issue:

    • Space

    • Time

      • Josh: metadata overhead implies that we might not gain

        much time

  • Symlink:

    • Petr: could be useful for publication and possible to

      explain

    • Publication workflow: locking file in “unix” sense to avoid

      deletion (cf. Thomas Zobel)

  • Delete:

    • Way to indicate what can be deleted, what is symlinked: no
  • Simon: annotation on the copied dataset so the user has a way to

    find out.

• SA:

  • eel/octopus still alive due to tied jenkins jobs

• Glencoe

  • Change from 5.6.1 custom build to be migrated to OS

  • Melissa: issue with processed data generating non normalized

    floating data in Tiff

    • Zoom level scaling not testing. Iviewer not using it, it is

      doing its own stuff

    • Few more steps to do

    • Existing Doc to be improved

• Community

  • Shutting down help.o.org -> redirecting to guide

  • Pdf will still be available under

    https://downloads.openmicroscopy.org/help/

  • Jason:

    • Community meeting output:

      • Start global discussion on different issues. Jason will

        start to organise things while taking time off away from zoom.

    • Ome face masks coming out soon

1. AOB

(5 min. max; tech. Discussion should be highlighted to relevant people and rescheduled)