centWave warning of not centroid mode data

May 02, 2012, 08:09:21 PM

The orbitrap .raw files were converted by ReAdW --mzXML -c input output

xcmsSet(mzXML.files, snames = snames, sclass = sclass, method = 'centWave', ppm = 10, peakwidth = c(20, 50), snthresh = 10, prefilter = c(3, 1000), integrate = 1, mzdiff = -0.001, verbose.columns = TRUE, fitgauss = FALSE)

Out of 167 samples, 15 gave the the following waring
In .local(object, ...) :
It looks like this file is in profile mode. centWave can process only centroid mode data !

What are the possible reasons?
What are the consequences on the following steps?
I found that retcor didn't work with the following error message: (after group)
"No peak groups found for retention time correction"

Details:
> xset <- group(xset)
262 325 387 450 512 575 637 700 762 825 887 950 1012 1075 1137 1200
> xset.retcor <- retcor(xset, family = 'symmetric', plottype = 'mdevden')
Error in .local(object, ...) :
No peak groups found for retention time correction
> traceback()
8: stop("No peak groups found for retention time correction")
7: .local(object, ...)
6: retcor.peakgroups(object, ...)
5: retcor.peakgroups(object, ...)
4: do.call(method, alist(object, ...))
3: .local(object, ...)
2: retcor(xset, family = "symmetric", plottype = "mdevden")
1: retcor(xset, family = "symmetric", plottype = "mdevden")
> dim(xset@groups)
[1] 2990 9

We can see that the groups element is not empty.

Any ideas? Many thanks.

Bowen

Re: centWave warning of not centroid mode data

Reply #1 – May 03, 2012, 07:34:42 AM

Quote from: "bowenli37"

In .local(object, ...) :
It looks like this file is in profile mode. centWave can process only centroid mode data !

This it is just a warning from a heuristic function to detect profile data.
It can be ignored if your samples are in centroid mode.

Quote

No peak groups found for retention time correction

This can happen with very large sample set, although I would expect that 2990 features should be enough to find at least one.
Are the group parameters (defaults) suitable for your setup?

You could also try the orbiwarp method,

Code: [Select]

retcor(method="obiwarp")

which does not require specific peak groups.

Carsten