https://nf-co.re logo
Join Slack
Powered by
I don't remember but has there been done any work ...
# references
n
I don't remember but has there been done any work on an actual references plugin? Strict syntax has changed some configuration parsing which makes the current way with 
params.genomes
less easy to use. It would be better if we could just move this logic to a plugin and resolve everything on the workflow level instead
m
tell me more
n
I don't have any specifics as I'm still figuring it out, but I had some issues when converting our preprocessing pipeline to strict syntax. E.g. the test profile doesn't use it's specified 
genomes
map anymore and instead tries to use the full genomes set which is a bit annoying
a
That seems more like an issue of profile order/params declaration with the strict syntax/v2 parser?
n
It doesn't work no matter the order I use 😕
m
I’d love a nice way to decentralize the reference handling, perhaps even coupled with a dedicated references pipeline that feeds into the plugin 😄
🙌 1
n
But anyways using 
params.genomes
is not ideal
m
a
Hmm do you have a minimal working example of that kind of failure?
n
I'll try and whip something up if I find some time 😛
m
coupled with a dedicated references pipeline
@Matthias De Smet this is exactly what this pipeline does (and this channel is for)
m
@mashehu, I know about the pipeline, just hinting about coupling the plugin with the pipeline output
👍🏻 1
a
For reference Nicolas this is is the sort of order issue I am talking about.
n
Aha okay I'll try moving the profiles blocks down
a
I looked briefly at your old 2.0.5 code (assume this was pre-strict syntax) and I don’t think it makes sense that it’s an order issue but you never know! I think there are definitely some edge cases with the new syntax and parser so even if there’s a better solution it would be good to work out what the issue is
n
The PR I'm having these issues with is this one: https://github.com/nf-cmgg/preprocessing/pull/167
Putting the profiles at the end is not solving the issue sadly 😕
The main problem here is that a custom igenomes config file is used in the test profile. Somehow it's not able to correctly read that one
a
I wonder if it is due to the 
map
param. I know Ben previously said that there is no attempt at merging, but perhaps map params in profiles do not properly overwrite as one would expect?
If that’s the case then I think this is an issue of wider significance
n
Yeah that's probably a factor here. In this case I would want it to overwrite the whole genomes parameter though
a
how do you enable static types? Do you have to add it to every file?
n
You can run the script converter of the vscode plugin on the main.nf file. This is not yet full static typing as that feature is not stable yet
I only typed the parameters here
a
But I don’t need to set a feature flag to test this?
OK I have tried several things with a very minimal example and I can’t seem to make it fail to pick up the overwritten map.
n
Nope, no feature flag needed.
Strange... I'll also give it a try later today
m
Hi, sorry I missed that, for now, I put things in the #C08TKFDUNHG plugin
n
aha thanks! I'll take a look at that soon 😛
Open in Slack
PreviousNext
Powered by