Problem with plots in the groupChromPeaks tool

[melpetera]

Hi,

The PDF output from the groupChromPeaks tool (XCMS suit) when using the PeakDensity method does not respond to the maxFeature parameter, and is not responding well to the minFraction (and potentially the minSamples) parameter.

Investigations led me to the plotChromPeakDensity function from the XCMS package, so I opened an issue on there package repository.

Following what may happen, a correction may be done on a future version of XCMS, but as we are still using XCMS 3 in the Galaxy tool, we may not benefit from the update for current versions. Consequently, would it be ok to provide a specific fix for the Galaxy module by kind of "duplicating the corrected version of the function" in lib.r for example, and replace the call to plotChromPeakDensity by the new "fixed" function?

I have no idea if it is a valid-enough solution, and if there are other alternatives that would be better.

Mélanie

[jorainer]

just sharing my opinion on that too - I think it would be better to fix the plot function upstream in xcms than having a local fix for it - also thinking in terms of maintainability and long-term support.

[melpetera]

Thank you @jorainer for your time. That would be generally the best thing indeed. Here what is tricky is that the Galaxy modules are currently working under an XCMS 3 version. So, if we want to benefit from the fix in xcms package directly, shouldn't we first upgrade to XCMS 4? Or is there another alternative?

To note, upgrading to XCMS 4 is anyway topical (and I don't think I am wrong saying it is also highly expected by part of the Galaxy users). Nonetheless we are currently short in manpower in the W4M team, so eventhough it is actually on our roadmap to upgrade, we will not be able to proceed shortly, for sure. So this is why I came up with the 'not-good-practice'-local-fix eventuality, in search for a compromise. But would this be an acceptable one? Not sure I am relevant to evaluate accurately that 😔

[jorainer]

Ah, yes, I totally understand!

Could you still please make a PR with your fix to the current xcms devel branch?

What you could do in the meantime - to also be future-safe is for example in the .R file that contains your fixed plotChromPeakDensity() method have a condition that redefines the method based on the installed version of xcms. Something like

if (packageVersion("xcms") < "4.6.0") {
    setMethod("plotChromPeakDensity", ...
}

and source that .R file after loading library(xcms). then, when the xcms version is lower than the fixed version you would re-define the method with your fixed version.

[jorainer]

Hi, again me - I did not mean to push/force you to do a PR. I meant it as a possibility for you to contribute - without much pressure. this would also be a possibility to give you credit for the very thorough explanation and description of the issue. But, as said, we can also fix the function - and in the meantime you can have the local fix in the W4M