I just tried to launch a run with Sarek 3.7.1, but got this:

N E X T F L O W  ~  version 25.04.3
Pulling nf-core/sarek ...
downloaded from <https://github.com/nf-core/sarek.git>
Launching `<https://github.com/nf-core/sarek>` [stupefied_brown] DSL2 - revision: 20f41d1ce8 [3.7.1]
Downloading plugin nf-schema@2.6.1
ERROR ~ Plugin nf-schema with version @2.6.1 does not exist in the repository

Nf-schema 2.6.1 is incompatible with your nextflow version. Should be fixed by updating nf

great, thanks for the reply. This is the latest version with Tower Enterprise, will need to pin a more recent version.

sorry about that

Another thing I realised: launching Sarek uses certain pre-filled fields such as

igenomes_base

or

snpeff

. Both these entries prohibit launching a workflow, even if one does not use SNPeff or any iGenomes, and even even it

--igenomes_ignore

is set to

true

. I suspect that the underlying reason for this is that we use our own (sso secured) AWS credentials, which prevents access to freely accessible S3 locations. But it is very counter-intuitive that things you might not have specified or care about prevent you from successfully launching the workflow, wouldn’t you agree? I found that specifying a valid S3 bucket in conjunction with

--igenomes_ignore = true

also fails with this error:

--igenomes_base ...: could not validate file format of 's3://.../': Unable to marshall request to JSON: Key cannot be empty.

I have found that deleting the default values entirely is sufficient to make the workflow start. Is there a way to make this work, or possibly at least mention in the Information tooltip?

ah good point. hm let's think about if we can handle the defaults better / the igenomes overwrite. Ideally no data should be accessed if the igenomes ignored flag is set. Would you mind dumping this into an issue? Also is this a new issue or one that you saw in older sarek versions as well>

I haven’t specifically tested this with older versions, but I think it might be a newer issue. I also saw the same behaviour with the new Ribo-seq workflow (1.2), so the validation seems to take place regardless of the `true`/`false` flags, or in the

SNPeff

case regardless of whether the tool in the listed in

--tools [list]

. I suspect it not necessarily a Sarek specific issues but something more general. Will open an issue shortly.

good to know. Maybe something in the template changed then. Thank you thankyou

Many thanks for looking into this! https://github.com/nf-core/sarek/issues/2079

Excuse the quick AI work, but is there a reason to not just do https://github.com/nf-core/sarek/pull/2083?

I see no issue with this as we're anyhow checking the path in https://github.com/nf-core/sarek/blob/dev/subworkflows/local/annotation_cache_initialisation/main.nf

That seems better, you can control when and how the existence checks happen

I'd just like to merge this PR https://github.com/nf-core/sarek/pull/2080 first, to make sure we're capturing all WARNING message and not missing anything

Cool, I'll leave it with you. Thanks!

brilliant, thanks

I do however suspect there may be something else going on https://github.com/nextflow-io/nextflow/issues/6668