Skip to main content
Topic: Obscure error with findPeaks.matchedFilter (Read 3499 times) previous topic - next topic

Obscure error with findPeaks.matchedFilter

Dear all,
I'm using the latest xcms with R 3.0.2 on x86_64-suse-linux-gnu (64-bit).
I've encountered a problem with findPeaks.matchedFilter which I can't find anything about in this forum or indeed anywhere.
Funny thing is, the same code runs smoothly on Windows 7 / 64bit with R 2.15.3.

This is the error:

Fehler in doubleMatrix(num, length(zidx)) :
  das dims Argument enthält fehlende Werte
(sorry for the German. It translates to: "Error in doubleMatrix(num, length(zidx)): the dims argument contains missing values")

traceback()
14: .Call("DoubleMatrix", as.integer(nrow), as.integer(ncol), PACKAGE = "xcms")
13: doubleMatrix(num, length(zidx))
12: profFun(object@env$mz, object@env$intensity, object@scanindex,
        num, minmass, maxmass, FALSE, object@profparam)
11: `profStep<-`(`*tmp*`, value = 0.5)
10: `profStep<-`(`*tmp*`, value = 0.5)
9: split.xcmsRaw(object, f = keepidx)
8: split(object, f = keepidx)
7: .local(object, ...)
6: findPeaks.matchedFilter(xraw, fwhm = fuwihama, sigma = fuwihama/2.3548,
      max = maxpeaks, snthresh = snlevel, step = stepsize, steps = nofsteps,
      mzdiff = mzdifference)
5: findPeaks.matchedFilter(xraw, fwhm = fuwihama, sigma = fuwihama/2.3548,
      max = maxpeaks, snthresh = snlevel, step = stepsize, steps = nofsteps,
      mzdiff = mzdifference) at tmptDMo6q.R#404
4: eval(expr, envir, enclos)
3: eval(ei, envir)
2: withVisible(eval(ei, envir))
1: source("tmptDMo6q.R")

The mass spectrum (from LC/MS) is a cdf converted from ChemStation and loads fine:

> xraw
An "xcmsRaw" object with 314 mass spectra

Time range: 5.4-361.2 seconds (0.1-6 minutes)
Mass range: 321.1-1931.9 m/z
Intensity range: 150-75984

MSn data on  0  mass(es)
        with  0  MSn spectra
Profile method: binlin
Profile step: 0.5 m/z (3223 grid points from 321 to 1932 m/z)

Memory usage: 7.96 MB

The mass spectrum was recorded with a pretty high threshold so it looks kind of sparse. Some scans do not contain any data points at all.
This is the output of xraw@scanindex:

  [1]    0    0    0    0    1    3    5    7    10    12    14    18
 [13]    23    27    33    39    43    45    46    49    53    58    62    65
 [25]    67    67    68    69    70    71    71    71    71    71    71    71
 [37]    71    71    71    71    71    71    71    73    75    77    78    79
 [49]    80    80    81    81    81    81    81    81    81    84    89    96
 [61]  101  108  113  117  122  128  134  142  152  164  177  199
 [73]  232  278  340  413  481  537  586  633  681  729  779  836
......

My suspicion is that is has something to do with that. Just strange that Windows seems to ignore that.
Anyone any ideas how to patch this? I need the linux version since it runs on our webserver that processes mass spectra and outputs data via CGI's.
Any help highly appreciated. I hope I didn't leave anything out.
Thanks in advance,
Tobias