https://nf-co.re logo
Join Slack
Powered by
Hey Team, <@UNCU4LKJM> made you <a present>, and <...
# methylseq
j
Hey Team, @Felix Krueger made you a present, and I gift-wrapped it rusty
rusty 3
🎉 1
f
ooooohhhh! 🤩rusty
j
Happy to merge this Felix, or do you want to wait for other review?
f
Happy either way, keen to see it in action 🙂
j
OK, merged with your approval! It was mostly just module update fallout anyway. Other maintainers- yell at me if you think there was anything dodgy here and I'll fix it.
👍 1
f
Maybe I should also get more involved in maintaining modules/versions - there is already a v3.1.0 release, and I expect there may need to be more updates over the next weeks and months when people use it for real life work…
👍 1
j
Should be easy to bump versions now I've updated the surrounding pieces
f
shall I have a go at bumping, and send it your way for review?
j
Sure! Let me know if anything puzzles you
f
Claude says: methylseq → nf-core/modules bump — fully staged and waiting on one input from you: the two container strings from seqera.io/containers for bismark 3.1.0 (bioconda + conda-forge) — the Docker community.wave.seqera.io/library/bismark:3.1.0--<hash> line and the Singularity blob URL. Drop those here and I’ll finish it end-to-end (update all 7 modules → regenerate nf-test snapshots against the real container via Docker → lint → push to your fork → open the PR pinging Jon). The clone + branch bump/bismark-3.1.0 are ready in the scratchpad.
do I have to do this by hand or shall jean claude code van damme figure it out?
So I can get the authoritative URI programmatically.: OK, it can do it
j
I think the new modules tooling should be able to do this for you now: https://nfcore.slack.com/archives/CE5LG7WMB/p1782311029429749
I have a claude skill I've been using for it, but it's probably superfluous now
f
j
You need to run commands lik 
nf-test .. --update-snapshot
. Though it looks like there are no changes except the version so you could probably cheat and just do a 
sed
to replace 
3.0.0
with 
3.1.0
I'm not actually sure if we're committing the lockfiles these days, I've asked the maintainers
Looks good now!
p
Note that it'd be nice to switch to using the single binary + subcommands once that's released
f
A PR for the latest version is out there now: github.com/nf-core/methylseq/pull/614 - but yea I think once this is proven working a few adjustments are in order as the new version has different recommendations.. E.g. the new lever to manage speed is going to be 
-p
and no longer 
--parallel/multicore
. Also the memory and CPU allocation will need to be adjusted.
I have tried to pick the best, fastest and most memory frugal modes and put them on this page of recommended practices
❤️ 1
This version bump switch to the Rust version, and should be a drop-in replacement. I will prepare another issue + PR for changing the Bismark route of the workflow fundamentally, i.e. change to single-pass processing against the e combined genome index using just 
-p
for speed. But this belongs in a different PR as it changes the default alignment behaviour (if only slightly). The other, sometimes dramatic, speed and efficiency gains can already be had in PR #614
Can you please look at github.com/nf-core/methylseq/issues/615 to see if you have an opinion on changing the Bismark default. behaviour going forward? I have a clear preference, but don’t want to bias anyone. @Sateesh_Peri @Phil Ewels @Jonathan Manning @Edmund. Once we’ve got consensus, we can go ahead and change the modus operandi accordingly. Cheers
💬 1
👀 1
2 Views
Open in Slack
PreviousNext
Powered by