Passing external arguments to DIA-NN workflow

[Lukas-Neuhauser-HMS]

We are using the DIA-NN based workflow with Bruker .d files. The pipeline runs on AWS with Nextflow 25.04.7 Quantms 1.6.0 and DIA-NN enterprise 2.2.0.

We are passing external arguments via ext.args = ... to individual workflow steps.

Currently, we sometimes want to pass --var-mod UniMod:35,15.994915,M --fixed-mod UniMod:-1,296.18,C. In our current testing we pass this to all 4 DIA-NN steps. Hence, to PRELIMINARY_ANALYSIS, ASSEMBLE_EMPIRICAL_LIBRARY, INDIVIDUAL_ANALYSIS, FINAL_QUANTIFICATION

Can we expect any unwanted side effects of passing the parameters to all 4 steps? Or is it fine to always just pass those extra args to all 4 steps and they will only take effect on the steps where they belong?

Can you give us an indication to which steps the parameters need to be passed and where they are simply ignored?

Thank you already

@ypriverol

[jpfeuffer]

Hi Lukas,

question: Why don't you use the real Unimod mods for this and use the parameters exposed for this in the SDRF instead?
Your second fixed mod is probably DBIA?

This is how to specify mods in SDRF

If you still want to pass your own arguments, you can just check in the source code, where we usually add modifications.
It looks like we do this by creating a global diann config in the beginning:
https://github.com/bigbio/quantms/blob/master/modules/local/diann/generate_cfg/main.nf

[jpfeuffer]

Unfortunately I see that this step does not take ext.args, yet :(
#641

[Lukas-Neuhauser-HMS]

Thank you for the clarifications @jpfeuffer. We will look into using the SDRF for mods.

I have one more related question. We sometimes want to pass --no-peptidoforms. Again we pass it as an external argument and currently we pass it to PRELIMINARY_ANALYSIS, INDIVIDUAL_ANALYSIS.

Is that the way to go or should we again simply pass it to all steps although it does not apply everywhere?

[ypriverol]

@Lukas-Neuhauser-HMS If you want to pass it as a global parameter to all, you can do a PR to dev and support it. I do like the idea of supporting more and more parameters from DIANN outside the "standard" way of running. Most of this parameters could be disabled by default and enabled if needed.

Another interesting topic is how you can pass custom parameters to multiple modules rather to one by one, is that even possible?

[jpfeuffer]

Of course. https://nextflow.io/docs/latest/config.html#selector-expressions
But I think he is asking from a DIANN standpoint. In which steps does DIANN even accept these parameters?!

[jpfeuffer]

I think one just needs to check "--help" of the diann commands to see.

[Lukas-Neuhauser-HMS]

Exactly. Since Quantms decomposes DIA-NN into 4 individual steps (commands). We are wondering to which one of the steps, which arguments should be passed.

Unfortunately, the documentation does not make that clear and also Vadim did not have a clear answer about that.

So it boils down to two basic questions.

To which one of the 4 steps which DIA-NN arguments can be passed. Where are they accepted?

If we do not have an answer. Is it fine to pass parameters to all steps, even to the ones where they are not accepted, without side effects?

[Lukas-Neuhauser-HMS]

I can of course create a PR to allow for passing additional args through the normal quantms mechanism (not as external args) but for that I would need to be sure to which steps I should apply the arguments.