[ome-devel] Genomewide bioinformatics data entry into OMERO with missing images.

[Josh Moore]

Migrating this discussion to the mailing list. Discussion inline. ~J.

> On 7 Jun 2016, at 07:49, Josh Moore <josh at glencoesoftware.com> wrote:
>
> Hi Ilan,
>
> some quick responses inline. In deference to our community, I'd
> suggest that we take any follow ups to the mailing lists and/or forums
> so they can get involved as well.
>
>
> On Thu, Jun 2, 2016 at 11:44 PM, Jason Swedlow (Staff)
> <j.r.swedlow at dundee.ac.uk> wrote:
>
> From: Ilan Davis <ilan.davis at bioch.ox.ac.uk>
> Date: Thursday, 2 June 2016 21:58
> To: Jason Swedlow <j.r.swedlow at dundee.ac.uk>, Darragh Ennis
> <darragh.ennis at bioch.ox.ac.uk>, Joshua Titlow <joshua.titlow at bioch.ox.ac.uk>
>
>
> ...snip...

> 3) Perhaps potentially most revolutionary for us. We would like to enter
> collections of protein trap genes that we will screen (all of or maybe only
> a subset) into OMERO, before we actually have any images associated with
> them. They could be identified as a gene in a tag for example. The reason
> for this is that we can then create intersects with various kinds of data
> (as I described in my talk) as a means of selecting priorities for which of
> the genotypes to image first and perhaps the kinds of imaging to do on
> subsections of the collections of gene traps (e.g. depending on where the
> gene is already known to be expressed). OMERO now turns out to be a very
> convenient place to carry out such intersects and they will be automatically
> linked to the future images by default, rather than having to then bring in
> information from other sources with the danger of mistakes creeping in -
> such as errors in the gene annotations.
>
> a) Can one have a tagged object with no actual image? i.e. import a empty
> image with no pixel information? Or even import it with an image and then
> delete the image leaving the metadata?
>
>
> Yes to all the above. Additionally a tag may exist without being
> linked to anything.



On Thu, Jun 9, 2016 at 4:41 PM, Ilan Davis <ilan.davis at bioch.ox.ac.uk> wrote:
> Hi Josh
>
> That is really good to know - thanks for these quick and clear answers.
> Being able to use the tags and metadata without images associated with some
> of them will allow us to try some very cool things - entering genome wide bioinformatics data in
> the OMERO database (and then viewing it in Zegami) so that to use
> preexisting data as a way to classify and intersect with future data will help decide what to image and
> add to the database of images.  This is a really different way to do
> microscopy than what we are used to and a very effective way to make use of
> bioinformatic and other data from a wide range of sources in an integral
> manner to OMERO. It will take some major thought before we implement this, incase it slows
> the OMERO DB down drastically.
>
> Do any of you see any potential performance issues?  i.e. entering genome
> wide bioinformatics data into OMERO but only populating a very small sub set
> of the gene tags with images.
> e.g. an OMERO DB with 13,000 gene tags and IDs but only say 1000 of the
> genes having associated images.

I don't know of any particular performance issues.


> One would need a way to show/hide
> information where no images exist for the particular tag.
> What would be the most efficient way of doing that?  Can one populate a key
> value pair automatically with image present / absent automatically?
> Any other ideas?

We'll most likely hit workflow limitations first. At that point, it'll
be best to have someone on your side who can prototype something (a la
webtagging) get it to users and relay performance / platform issues to
us.


> I did not add this to the list / Forum because I do not know which one to
> use or how. But I am happy for this to be placed by anyone else there for
> discussion.

Anyone else trying something similar? Thoughts?


> Regards
> Ilan


~Josh