My problem is that the features identified by XCMS are off with around 0.02Da compared to raw files (.d) accessed by Bruker's Data Analysis 4.2.
What could be the explanation for this?
I am using RStudio (Version 0.98.994), XCMS (version 1.38.0).
I have a HPLC-qTOF (Bruker) which produces .d files. I have converted these files into .mzXML files by CompassXport 3.0.
Following I run
library(xcms)
xset <- xcmsSet(myDir, ppm = 30, peakwidth = c(10,60), method = "centWave", snthresh = 10, noise=100, prefilter = c(4,1000),
mzCenterFun = "wMean", integrate = 1, mzdiff=-0.01, fitgauss = TRUE, profparam = list("step" = 0.01))
xset1 <- group(xset, bw = 30, minfrac = 0.5, minsamp = 1, mzwid = 0.025, max = 50)
xset2 <- retcor(xset1, missing = 1, extra = 1, smooth = "loess", span = 0.2, family = "symmetric")
xset3 <- group(xset2, bw = 30, minfrac = 0.5, minsamp = 1, mzwid = 0.025, max = 50)
xset4 <- fillPeaks.chrom(xset3)
reporttab <- diffreport(xset4, "A", "B", "Outputfile2", 11, metlin = 0.15, h = 480, w = 640)
xset5 <- peakTable(xset4, filebase="name")
Any help is really appreciated!
Check the mzXML files for example in mzMine. It might be that the files have been converted uncalibrated:
ref: http://http://link.springer.com/article/10.1007/s00216-013-6954-6
Thanks. I had a look at mzXML files in mzMINE and it seems to be a problem with the conversion from .d to mzXML.
I had a look in the registry key, but I can't see any wrong settings there... So I'm still lost.
I tried to export to mzXML files manually from Bruker's DataAnalysis, thereby not using CompassXport. (The disadvantage is that you can only export one file at a time, so it takes more time compared to CompassXport)
However, by doing so I fixed the problems and got the 'real' mz values after running XCMS.
So the problem is the conversion by the CompassXport program.