Re: cwl - ok for me to document my progress on wiki.d.o?
Hello,
على الأحد 14 شباط 2016 23:49، كتب Andreas Tille:
> On Sun, Feb 14, 2016 at 09:27:40PM -0800, Afif Elghraoui wrote:
>>> So it should be added not only to med-bio but also to med-imaging,
>>> right?
>>
>> Shouldn't it rather be put into the Debian Science Tools task [1]
>> together with snakemake, make, makepp, and other workflow management
>> tools? It might make more sense to keep the
>> biology/imaging/other-specialized tasks focused more on directly-related
>> software.
>
> science/tasks(master) $ git diff HEAD^
> diff --git a/tasks/tools b/tasks/tools
> index ab0464c..388cdee 100644
> --- a/tasks/tools
> +++ b/tasks/tools
> @@ -18,3 +18,5 @@ Depends: xoscope
> Suggests: libmemtailor-dev
>
> Depends: python-h5py | python3-h5py
> +
> +Depends: snakemake, cwl
>
:) So is there no problem to edit tasks pages for other blends to add
packages we maintain? I always wondered why Debian Med didn't maintain
the science-bio metapackage directly. I thought that med-bio was created
and added as a dependency to science-bio just so that it could be
managed directly by Debian Med. I never got around asking about that, I
guess until now. Was the reason for this just historical?
>
> I do not see any reason to add quite generic tools like make & makepp.
>
As you wish, though they do have relevance for scientific research. I
personally use Make for all my bioinformatics pipelining. It's
especially nice since you can change the SHELL/.SHELLFLAGS for different
targets. If you wanted, you can have one rule execute in R, one in
Bourne shell, another in SQL, and others in Python, Perl, and so on. You
might think of snakemake almost as running a regular GNU Makefile that
sets `SHELL=python` and declares `.ONESHELL`.
> However, despite the fact that I think these tools are well placed in
> science-tools tasks are not exclusive. A task should give a hint to a
> user looking at the task matching the own field of work. So there is no
> point in putting snakemake and cwl into only science-tools. We could
> decide that it is not that important for med-bio / med-imaging and just
> use a "Suggests" relation.
>
Or should med-bio/med-imaging just suggest science-tools itself? I think
the other things in there may also be of interest and we wouldn't need
to add Suggests one at a time. I have no problem either way; just a
suggestion.
Thanks and regards
Afif
--
Afif Elghraoui | عفيف الغراوي
http://afif.ghraoui.name
Reply to: