featureGrouping error

[stolltho]

patRoon 1.2.0 xcms 3.8.2

Hi Rick I was analysing 350 samples (~70GB) w/ xcms3. During feature grouping (method: PeakDensity) I received this error message: Error in result_bind(res@ptr, params): long vectors not supported yet: ../include/Rinlinedfuns.h:522

Is this an xcms or patRoon issue and how can it be fixed?

Cheers, Thomas

[rickhelmus]

Hello,

That's quite a bit of data :-)

To be honest I never tested a large batch like that myself, as we are usually working with smaller ones. As these larger batches seem fairly common in metabolomics, I would expect XCMS could handle them though...

Anyway, perhaps you can try a few things to get more details:

• Run a small subset to see if that works and therefore confirming it is indeed a size issue.
• Disable retention time alignment (pass rtalign=FALSE)
• Disable caching, eg fGroups <- withOpt(cache.mode="none", groupFeatures(...))

If you want and it is possible you can also share the data so I can look at it myself.

Thanks, Rick

[stolltho]

Hi Rick

Thanks for the suggestions. After disabling caching the analysis finished successfully.

Cheers, Thomas

[rickhelmus]

Hi Thomas,

Glad to hear this can work as a work-around... but it doesn't really fix anything ;-)

Could you do me a favour and see if this is related to the batch size? For instance:

fGroupsSub <- groupFeatures(fList[1:5], "xcms3") # only group features from first 5 analyses

And try with increasing numbers of analyses.

Thanks, Rick

[stolltho]

Hi Rick

I recently analyzed 350 sample @ 5 min run time (data size 24 GB) and the code finished fine. This time the run time was 20 min (data size 90 GB) and it returned this error which only went away after the cache work-around. I guess it is a batch size problem.

Cheers, Thomas