These are chat archives for nextflow-io/nextflow

15th
Mar 2016
Gijs Molenaar
@gijzelaerr
Mar 15 2016 08:07
Hi all. I wanted to bring under your attention this project i’m working on called kliko http://kliko.readthedocs.org/. We created kliko to create a bit more structure into using containers for scientific computing by formalising the input, output and parameters. Although I still need to investigate the usefulness of combining kliko and nextflow people here may be interested in the progress or have suggestions.
Paolo Di Tommaso
@pditommaso
Mar 15 2016 08:24
@gijzelaerr Thanks for sharing this. How do you orchestrate/compose the execution of multiple containers ?
Gijs Molenaar
@gijzelaerr
Mar 15 2016 08:24
I imagine nextflow comes to play there
Paolo Di Tommaso
@pditommaso
Mar 15 2016 08:24
ah, sounds good.
Gijs Molenaar
@gijzelaerr
Mar 15 2016 08:24
we now use this for scheduling single parameterized compute containers
but there is nothing stopping people from chaining them.
I try to enforce no side effects, so it fits the functional model of nextflow
Paolo Di Tommaso
@pditommaso
Mar 15 2016 08:26
Yes, it does. However the constraint on having a /input and /output folders in the container could be a problem if you want to use nextflow.
Gijs Molenaar
@gijzelaerr
Mar 15 2016 08:26
why is that? Sorry i have not had the time to fully dive into nextflow yet but on paper it looks like a logical thing to use
Paolo Di Tommaso
@pditommaso
Mar 15 2016 08:28
because nxf just mount the input files path on an identical path in the container. The idea is allow containers usage in a transparent manner.
Gijs Molenaar
@gijzelaerr
Mar 15 2016 08:30
ok I think this means I just need to start playing in trying to combine kliko and nextflow and see what and what doesn’t work
Paolo Di Tommaso
@pditommaso
Mar 15 2016 08:30
:+1: