Re: Metadata exchange with the Tools Platform Ecosystem
Hi again,
On 11.11.20 17:39, Hervé Ménager wrote:
> I can only agree with Matus, this is a very nice and super motivating
> answer, thanks a lot!
> We'll investigate how all of that can be delivered in a near future
> hopefully :)
My personal ambition is a substitute for our task pages. But we need the
EDAM annotation for that. Let us see where our ambitions to tackle the
pandemic carries us. EDAM-annotating just these packages and then
presenting the worksflows with it should be an interesting excercise.
> On Wed, Nov 11, 2020 at 5:28 PM Matus Kalas <Matus.Kalas@uib.no
> <mailto:Matus.Kalas@uib.no>> wrote:
>
>
> Giant thanks for your quick and enthusiastic answer!
>
> On 2020-11-11 16:59, Steffen Möller wrote:
> > Hello Hervé, hello Matúš,
> >
> > On 11.11.20 16:36, Hervé Ménager wrote:
> >> Hello Debian-ers,
> >> We (ELIXIR Tools Platform) have been working a lot on the Tools
> >> Platform lately. One of the major contributors is Debian Med, and
> >> collecting the package metadata from you will soon enable:
> >> - cross-linking between e.g. bio.tools and Debian Med packages
> >> - cross-validation and enrichment of metadata.
> >
> > How cool is that!
> >
> > I just checked https://bio.tools/clustalo
> <https://bio.tools/clustalo> and found the "software
> > package" link to Debian's tracker. Great!
> >
> >> Speaking of which, our current setup is very convenient for us:
> we can
> >> update Debian metadata at any time, and use it to produce
> better tool
> >> descriptions. *But*, one thing which is unclear is how we can
> >> contribute back some metadata to your packages. Would there be any
> >> kind of interest on your side in e.g. opening Merge Requests on
> salsa
> >> when some metadata can use some update? If so, should our
> system open
> >> these MRs automatically or semi-automatically (assuming we can
> define
> >> precisely when a metadata difference mandates a correction on the
> >> Debian side)?
> >
> > You personally have access to salsa.debian.org/med-team
> <http://salsa.debian.org/med-team> and can go for
> > anything exceptional without further delay.
> >
> > You can also prepare and auto-prepare (!) pull requests of whatever
> > nature these may be for all packages that are on salsa.
> >
> > For packages in Debian Med, fixing smallish bugs, like
> > adding/correcting
> > the bio.tools reference I think you can just do them.
>
> Great!
>
> We could correct the bio.tools reference for those tools|packages
> that
> have a link to Debian's tracker, but those would expectedly have a
> valid
> bio.tools ID in Debian already :) Or am I wrong?
>
If I recall correctly then there we are few packages were we had
mismatches with the cardinality. The assignment is currently to source
packages, but if a tool provides multiple binaries then we may be in
trouble if there are multiple bio.tools entries for that.
But - go ahead. Yes. Sounds like a good start and maybe you can check
for multiple assignments to the same tracker URL, first.
>
> > A seed for the edam annotation would be good, which then the
> individual
> > maintainers
> > extend, by chance.
>
> This is an excellent idea!
>
Right after getting the references synced, this is very much what I
would like to see going.
I manually went to a few packages of the "Virus" list on that
spreadsheet. Invitation to edit just went out.
Could you please investigate why there is no ref for
https://bio.tools/edger ? The naming scheme is
r-(cran|bioc|other)-lowercasepackagename. But URLs etc should be
correctly set. I just checked
https://salsa.debian.org/r-pkg-team/r-bioc-edger which has the right ref
to bio.tools (without me now having edited it :o) ). Hm. Maybe that is
not exported in the json file? But I checked
https://salsa.debian.org/blends-team/med/-/blob/master/tasks/bio and it
is listed there.
Do you have statistics on the number of entries compared, %matched,
%links to Debian added, or whatever seems appropriate? Also, the number
of lines you add from bio.tools to Debian should somehow be metered,
preferably over time. I know, just the typical bloat for whatever
manuscript you likely plan to write, but nonetheless nice to read :)
Best,
Steffen
>
> > I do not think I would in an automated way update
> > package descriptions. And the URLs should also be checked manually.
> > Even
> > if you have the correct newer one, the one that is listed is
> likely the
> > one where the software was downloaded from and it identifies the
> > sources, too.
>
> Hmm, good points.
>
> Maybe we'll have to be a bit careful about the versions of the source
> pkg in Debian and those that a particular bio.tools record is
> valid for
> (N.B. such annotation is optional in bio.tools). Such check, where
> possible, may apply to generating the edam seed, too.
>
> > More important in that respect is that the debian/watch
> > file is updated so the maintainer is informed about the updates.
>
> Ok, thanks for the reminder, I see this is crucial.
>
> >
> > As a start, I think a mere web page with lists of changes that
> you want
> > to feed back would be nice so we can think along.
>
> Indeed, I think this is the way to start, before we dive into complex
> functionality.
>
> >
> > Thank you both!
> >
> > Steffen (has added/updated already three bio.tools references
> today :o)
> > )
>
> So super cool, first place medal Steffen!
>
> Thanks so much,
> Matus
>
Reply to: