[Bio-packaging] GNU Guix update

Pjotr Prins pjotr.public66 at thebird.nl
Sun Jul 12 09:38:00 UTC 2015

The Common Workflow Language (CWL) is going to be key to rolling out
pipelines in bioinformatics. As it is mostly a static definition of
workflow containing input -> procedure -> output, it looks to me that
it can work well in conjunction with Guix expressions. Basically CWL
is like running a (distributed) gcc compile on inputs with outputs. 

So, defining a reproducible data analysis can be a Guix expression
with the CWL as a 'compiler'.

The CWL specification itself is mostly JSON. To create a more conscise
representation we could introduce S-expressions to generate this JSON.
CWL is a bit like RDF - it is mostly a standard description of a
typical workflow in bioinformatics. Workflow runners honour this
standard. And docker does *not* need to be part of that solution. The
solution can be pure Guix (and Guix does distributed builds already).

Using Guix as a cwl-runner (on a local build farm) is an interesting
thought for reproducible data analysis.


More information about the bio-packaging mailing list