[ome-devel] IDR Paper Out!

[Jason Swedlow (Staff)]

Dear All


Hope you all are having a good summer (in the Northern hemisphere) or a cool and relaxing winter (in the Southern Hemisphere).


Many of you will be aware of our Image Data Resource project (IDR; http://idr.openmicroscopy.org/), a collaborative project between the Dundee OME team, Alvis Brazma’s group at EMBL-EBI and Rafael Carazo-Salas’ group at Bristol.  The paper that describes this work is (finally!!!) fully published at Nature Methods (http://www.nature.com/nmeth/journal/v14/n8/abs/nmeth.4326.html).  Note that we have deposited the author pre-print for open access at PubMedCentral at https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5536224/. An editorial that discusses some of the implications was published by Nature Methods at http://www.nature.com/nmeth/journal/v14/n8/full/nmeth.4389.html.


The arguments for sharing and integrating image data will be familiar to the OME community. In the IDR project, we want to promote the re-use of the image data and metadata we have received from the study authors.  We of course provide a browseable UI that exposes the metadata integration and curation (getting all this to work and perform was no small feat!). However, we have also made the backend database available for download, the full IDR application deployable by Ansible, and, with support from EMBL-EBI, the image data downloadable by Aspera. This should allow the community to access IDR data and technology in many different ways.  Community members can build their own IDRs and integrate their own data and analytic tools. We don’t believe there will be a single IDR, but rather hope to promote the routine development and use of integrated imaging resources in projects and institutions worldwide.


Our work raises several questions. Here are some we have heard, and our current answers, which might change as things evolve:


  1.  Which images should be included in the IDR?  Should a lab submit all the image data it has ever collected?  Clearly, we don’t publish our lab notebooks and there is little justification for publishing all recorded bioscience and biomedical imaging data. The IDR implements the concept of a “reference image”, which was defined in a white paper<http://www.eurobioimaging.eu/content-news/euro-bioimaging-elixir-image-data-strategy> published by Euro-BioImaging<http://eurobioimaging.eu/> and Elixir<https://www.elixir-europe.org/>. The IDR project’s submissions page<http://idr.openmicroscopy.org/about/submission.html> details our implementation of this concept. There’s no doubt the reference image concept will evolve, but this seems to be a useful strategy to get going.  If you have ideas about what would be a “reference image” or reference image dataset for you then we would love to hear your definition. (By the way, if you are an ML person, “reference image” equates to “images with great labels”.)


  1.  What about a standardised data submission template?  We’d love to standardise experimental, imaging, and analytic metadata, but that’s not yet practical. We have built a metadata template inspired by the ISA-TAB formats that we are using for IDR data submissions, and this has proven useful for several of our recent datasets.  Comments, suggestions and updates are always welcome. (Note to imaging system vendors: supporting open file formats would help us A LOT!!!!)


  1.  How much data can we store and host?  EMBL-EBI is providing hosting resources for IDR through their Embassy cloud resource, so they’ll have to answer that one.  So far, they just keeping giving us more…


  1.  How sustainable is IDR? We think a key part of this issue is demonstrating the scientific value of building and maintaining image data  resources— that’s where IDR’s focus will be for the next year or so. We are certainly working hard to secure long-term funding for IDR resource and technology development. We’re very grateful to the BBSRC for funding work on the IDR to this point.

If you think IDR is important, please do contact the BBSRC, Wellcome Trust, NIH and others and tell them what you think!!!  If you have used the images or metadata in IDR, let us know-- examples and results will help our case.


  1.  Is the IDR a production resource? We’re proud of what we’ve achieved with the IDR, and look forward to continuing to grow the number and expand the range of datasets it holds.   We have a lot more technical development to do, and we have only begun to explore the possibilities for analysing and integrating IDR data. IDR should therefore not be considered a full production resource, but is still a proof of concept or prototype. Nonetheless, we expect the images that IDR holds to be available long term-- that’s why we have placed IDR at EMBL-EBI. It’s early days and we’ll have to see how this field and technology evolves.


  1.  How will you federate multiple IDRs? Good question. We applied for funding for the infrastructure to do this work twice— the reviews were great, but the panels decided not to fund the work. The feedback indicated that the panels want us to solve the sustainability problem without grant funding. How we were supposed to do this wasn't specified.


As always, we welcome suggestions, comments, feedback, criticism and most definitely usage and development of these tools.


Cheers,


Jason


The University of Dundee is a registered Scottish Charity, No: SC015096
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openmicroscopy.org.uk/pipermail/ome-devel/attachments/20170802/1c757e64/attachment.html>