Re: Do we benefit from a Debian Med machine to run our workflows?
On Tue, Jan 22, 2019 at 10:32:30AM +0100, Steffen Möller wrote:
> Hello,
>
> bcbio just arrived in the New Queue. Somewhat embarrassingly it does not
> install, yet, since python3-seqcluster is still missing who have their
> Python3 fixes in their devel branch. I am hoping for an upcoming official
> release or will package that branch.
We should use git mode in watch file if HEAD works for our purpose.
BTW, I do not think its a good idea to upload a package before its
pre-depends are fulfilled.
> Either way, we are about to have our first workflow package in our
> distribution. Now, I would like to test a bunch of research directions bcbio
> supports. And tests shall be executed "continuously" (repeatedly) as we
> update our packages. We have Debian's CI infrastructure - but tests run for
> several hours and our CI infrastructure may not be perfectly suited for
> that? Very much so since we are likely to desire access to the machine to
> inspect what is going wrong? And not everyone with a genuine interest is a
> DD?
>
> Admittedly, what I am more after than "integration testing" is "knowledge
> integration". A imagine a site to which Debian Med can log in and adjust
> parameters (maybe even in some automated fashion) to have the same dataset
> analysed again and again as the reference databases/software underneath
> matures/develops, preferably with some gold standard as a reference or some
> easy access to the same material for labs around the world to prove stuff -
> this may turn out to be interesting. And when there alternative workflows
> arriving in our distro, like with GATK, this may become even more
> interesting. Setup and maintenance would need to be completely automated -
> we just don't have time for anything else.
>
> I was offered a 2x6-core 64GB (I know) server with rack space. It has a 2TB
> drive and slots to add more. That we could play with.
>
> Comments?
Just do it.
Kind regards
Andreas.
--
http://fam-tille.de
Reply to: