Re: TR: Debian has RRID SCR_006638
Hi Dylan,
On 4/25/18 9:43 AM, Dylan Aïssi wrote:
> Hi Fabien,
>
> 2018-04-25 8:37 GMT+02:00 Fabien Pichon <fabien.pichon@omictools.com>:
>> If what Dylan proposed is convenient for a majority of persons (and if the
>> list and script he sent are correct) I can have something ready this week or
>> the next one.
> Of course, they are not correct :-) I will have another look today but
> it should be better to use UDD instead and I have to learn how to use
> it.
>
> But, what do you need exactly? Only new packages (first upload into
> Debian) or updated packages (I guess it is the first one).
> I will have to manually cure the list because some packages could be
> not related to omics... I am not sure if it is useful to highlight
> them on your blog.
;) I keep asking myself the same question and do not really have an answer.
The list of additions is short, at least once we are past this initial
phase it
will not be longer than what goes through our new queue. And the reader
of the blog with a particular scientific interest will not know why that
list
was created in the first place.
So, I think I prefer a general blog entry describing the link from OMICtools
to Debian (and Conda). That should state that software for which there
is a Debian (Conda) package is indicated as such in OMICtools and how to get
the package to work in a local installation. That blog entry should appear
once the filter for Debian (Conda) packages in OMICtools is implemented.
After all, we use OMICtools for the exact reason to get lists of software
and we should not create our own. The reason for us not focus on Debian
alone,
i.e. that we also care for Conda, is how I interpret our DFSG which states
that we should not have anything for Debian alone.
Fabien attended our Sprint in Barcelona. We also discussed additional
features like users of OMICtools asking for a Debian (Conda) package
to be created andthe creation of an RFP. Once these features are implemented
I am confident OMICtools will blog about them.
>From my side, I think we should stop thinking too much about individual
packages. It is workflows that count since the workflows are which map
to problems the users have, not so much the individual software. So,
Debian Med should establish these workflows, preferably with the CWL,
and in my mind these CWL files plus documentation (literature references!)
and references to test data are then Debian packages in their own right.
Once we get any such package in our distribution it makes decent sense
to blog about it. And together the concept to treat a biological workflow
as a regular piece of software that runs through our continuous integration
testing should evolve into a scientific publication.
Debian Med should not do this alone. Take some NGS copy number variation
workflow. There are many ways to get there. Different platforms will differ
in the versions of software used or find a different tool to transcribe file
formats or have different default settings. We need some catalog to our side
to indicate these differences to help guiding the users to the platform of
their choice or to a virtualised mix of platforms.
Best,
Steffen
Reply to: