Proposal for 4D Nucleome Microscopy Metadata Guidelines based on the OME data model

Dear All

the 4D Nucleome Imaging Standards Working Group would like to solicit feedback on proposed guidelines we are putting forward for a metadata standard for describing:

  1. MICROSCOPE HARDWARE Specifications
  2. IMAGE ACQUISITION Settings
  3. OPTICAL, FIELD, EXCITATION POWER and WAVELENGTH, and DETECTOR CALIBRATION Procedures

used to document imaging experiments and foster their reproducibility.

The guidelines incorporate and update the OME data model and aim at increasing data fidelity, easing future analysis and facilitating objective comparison of different imaging datasets and imaging experiments setups.

The proposed 4DN-OME guidelines have the following distinctive features:

  1. Based on the premise that not all imaging experiments are “created equal” and that different reporting guidelines might be required depending upon the experimental intent, we propose a Tiered system of guidelines that specify what kind of metadata is required for different kind of imaging experiments.

  2. Attempt to transition the OME data model towards a clear set of community standards that could serve as a basis for an “Encode”-like standard for imaging.

  3. Help to guide different stakeholders, including manufacturers, experimental and computational biologists, peer reviewers, publishers and funding agencies in evaluating of the performance, quality and uncertainty associated with specific microscope setups.

The current version of the metadata model and tier system description can be found on GitHub at the following address:

Please feel free to contact us for any question, concern, comment, (constructive) criticism.

Cheers

Caterina Strambio De Castillia and David Grunwald

I have a few questions/comments:

  • Who is the intended target audience for implementation? Is this something that should be included into OMERO and/or output by the microscope control software?
    From my experience, this is too much information for microscope users to provide reliably unless it is exported for them by the microscope control software.
  • Who is the intended target user of this information?
    The information captured here may be useful for reproducing an experiment but, again based on my experience (re-)using microscopy data, I see little information of use to computational biologists. As examples of what computational biologists are more interested in, I can list:
    • capture in a standardized way what the entities imaged in each channel are.
    • get a proper, standardized description of controls vs treatments (i.e. is negative control an untreated condition or is it treated in a different way ?)
    • standardize spatial position of samples (e.g. should plate wells be identified a row and a column or by counting, starting from where ?)

For anyone following along, here’s the summary that I posted to https://lists.openmicroscopy.org.uk/pipermail/ome-devel/2019-May/004366.html

A week ago, Caterina kindly organized a (very early US-morning!) call for users and developers who had responded with interest to this proposal, including individuals from the 4D Nucleome Imaging Standards Working Group, German BioImaging, RIKEN, and the University of Dundee. A number of uses & benefits of the distinctive features she listed were discussed with the general consensus being that having a way to flexibly capture more metadata about image acquisition would support a large number of different groups, from vendors to bench scientists.

The primary issue raised, however, was the speed of the OME-XML schema release cycle as well as the necessary related code changes and database upgrades for developers and sysadmins. Norio presented his work on translating the OME-Model into RDF/OWL. The “open world assumption” of RDF/OWL means that a schema update is not needed for someone to make new and interesting statements about an experiment.

With this type of extension mechanism, the community (i.e. you) should be able to propose terms and values that you would like to capture within OME-XML / OME-TIFF, while still having the guarantee that as the model evolves, we will keep your data up-to-date. Developers should be able to build flexible GUIs for parsing, displaying, and editing these values. Curators and standards bodies should be able to cooperate to manage this expanding vocabulary.

At the moment, Caterina, Norio, & Co. are investigating revising the proposal to make use of RDF & OWL. At the same time, the extension framework will need to be specified for allowing the import and export of extended metadata within the OME ecosystem.

Various parts of this work will soon start showing up in the ome-model repository, but we look forward to your thoughts – from what elements other than you would like to extended to what vocabularies you could imagine yourself using.

Hi J-K,

I’d say optimally output by the control software, but then maintained as losslessly as possible into and back out of OMERO, Bio-Formats, OME-XML, etc.

All agreed. @Caterina can say more, but the impetus was extending <Instrument/> for quality control measurements. @norio.kobayashi was motivated by correlative concerns. And like I mentioned in my summary above, I definitely can see others (like oyu) being interested in extending OME’s <Experiment/> and beyond.

~J.