Search
Question: xcms and xcmsSet no longer works on my Mac
0
23 months ago by
nickpardikes0 wrote:

Greetings,

I have been using XCMS in R for chemometric analysis for the past few months. Just the last few days the xcmsSet function on my computer (MacPro 10.12.1) (R version 3.3.2 (2016-10-31)) no longer works. Everytime I try to run the peak picking process R stops responding. I have to force quit and restart the program. I have uninstalled R several times, as well as the packages.

I recently downloaded mzMine2 for Mac OS X and perhaps that has something to do with these issues. Having said that I deleted it and it still does not work.

I tried the same code (as shown below) on a PC and it had no problems.

It seems the issue is with my computer, or at least the versions of the packages and R that I have installed.

Has anyone had this issue before and is there someway I can start using xcms again.

Other functions such as:

plotChrom(xcmsRaw(junxml[3]),fitgauss=F, rtrange=c(0,1500))

work fine. So something is going on with xcmsSet().

Here is the code that no longer works (junxml are my .CDF files, I only have 18 of them)

xset=xcmsSet(junxml, step=0.5, profmethod="bin", fwhm=5, snthresh=2)

Here is my sessionInfo()

R version 3.3.2 (2016-10-31)
Platform: x86_64-apple-darwin13.4.0 (64-bit)
Running under: macOS Sierra 10.12.1

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

attached base packages:
[1] parallel  stats     graphics  grDevices utils     datasets  methods   base

other attached packages:
[1] xcms_1.50.0         Biobase_2.34.0      ProtGenerics_1.6.0  BiocGenerics_0.20.0 mzR_2.8.0           Rcpp_0.12.8

loaded via a namespace (and not attached):
[1] RANN_2.5               lattice_0.20-34        codetools_0.2-15       MASS_7.3-45            MassSpecWavelet_1.40.0 grid_3.3.2             plyr_1.8.4             stats4_3.3.2
[9] S4Vectors_0.12.0       Matrix_1.2-7.1         splines_3.3.2          BiocParallel_1.8.1     RColorBrewer_1.1-2     survival_2.39-5        multtest_2.30.0

Cheers

modified 23 months ago by Johannes Rainer1.3k • written 23 months ago by nickpardikes0
3
23 months ago by
Mike Smith2.9k
EMBL Heidelberg / de.NBI
Mike Smith2.9k wrote:

I've occasionally seen the combination of a MacBook Pro and BiocParallel in MultiCore mode exhibit this behaviour.  It looks like XCMS makes use of BiocParallel, so if you force it to run process things serially, does it improve?

library(BiocParallel)
register(SerialParam())

xset <- xcmsSet(junxml, step=0.5, profmethod="bin", fwhm=5, snthresh=2)

This worked! I wonder why it suddenly stopped working though? Do you think mzMine had anything to do with that? I know that it uses xcms for peak convolution. Thanks a lot for your help.

1

I did a little more digging and I think you're probably encountering the same issue seen in bplapply doesn't work in the multicore mode on mac os x and BiocParallel not proceeding

My understanding of the discussion there is that some Apple OS update has made communication between processes  more difficult.  The suggestion is to run options(bphost="localhost") once, before you get to the part that is to be executed in parallel.

If that works for you, and you're able to process the files in parallel again, then posting a comment on one of those threads should bring it to the attention of the Bioconductor Core team.

I don't think it is related to mzMine.

Indeed, xcms was updated to use BiocParallel in Bioconductor release 3.4 to ease the parallel processing setup.

0
23 months ago by
Johannes Rainer1.3k
Italy
Johannes Rainer1.3k wrote:

This is indeed strange. I'm using xcms on a MacBook Pro without any problems. What happens if you call top -o cpu in a terminal - do you see R processes with a %CPU > 0? One side effect of using BiocParallel is that you don't see the progress of the computation, so it might well be that R is calculating.

What happens if you use the CDF files from the faahKO package? Do you get the same problem?

dir(system.file("cdf/KO", package = "faahKO"), full.names = TRUE)