Re: https://github.com/nf-core/covid19 - nextflow pipeline
Hi Steffen,
thanks a lot for your clarification.
Can we expect a list of links pointing to packages that are needed?
As I wrote in my other mail ftpmaster seems to be motivated. Currently
its us who need to deliver.
Kind regards
Andreas.
On Tue, Mar 24, 2020 at 03:37:22PM +0100, Steffen Möller wrote:
> Hi Tony,
>
> On 23.03.20 17:04, Tony Travis wrote:
> > On 23/03/2020 14:38, Steffen Möller wrote:
> > > Dear all,
> > >
> > > With the enormous difficulties to get bcbio into our distribution (a
> > > considderable list of sequential dependencies for nim, java libraries,
> > > golang-bio, ....sigh) I also had an eye on nf-core and their
> > > nextflow-based workflows. I think I uploaded the one or other package
> > > from their dependencies but, well, treated it more like an inspiration
> > > than a project. From what I oversee it is far easier to address than
> > > bcbio.
> > > [...]
> >
> > Hi, Steffen.
> >
> > I've just had a look at the "nf-core" repo and read this:
> >
> > "please only use Conda as a last resort"
> ;) You read something else elsewhere.
> >
> > They recommend running "nf-core" under Docker or Singularity, which I
> > believe would make it unsuitable for packaging as part of "med-bio".
> No, not at all. We get the packages into our distribution and there will
> be an additional recommendation. We just need to show that our packages
> are functional and they have done a nice job with their collection of
> test data. I (once upon a time) had an email exchange with upstream -
> they are very Debian friendly. And so are the folks behind conda and bcbio.
> >
> > I'm currently running "bcbio_nextgen" under Ubuntu-MATE 18.04 +
> > med-bio, but the developers advised me NOT to use the 'external'
> > versions of tools like "bwa" or "bowtie2" from Debian-Med but install
> > everything isolated from the 'system'.
> This avoid confusion. I'd recommend the same.
> >
> > I think this sort of application is more effort than it's worth to
> > package for "med-bio". However, I think that a bare-bones package
> > equivalent to bootstrapping "bioconda" using "miniconda" would be
> > worthwhile to provide a ready-to-use "bioconda" within Debian-Med.
>
> I concur. And this will happen.
>
> Anybody who is packaging for Debian is very likely to come up with
> patches for upstream. This is what also makes upstream stronger. The
> benefit of packaging after aims set by the nf-core community is that we
> get our act together towards something tangible (a set of proven
> workflows with test data with relevance for the ongoing pandemia) and to
> improve the packages upstream.
>
> Steffen
>
>
>
--
http://fam-tille.de
Reply to: