Quantifying centroided MS data with centWave

September 20, 2014, 12:38:56 PM

Hi all. This may be a fairly elementary question, but I am a relative newcomer to the nuts-and-bolts of quantifying MS data. The data that my lab collects is centroided before analysis. Since each data point is an intensity at a discrete m/z and not a continuous trace like you get with profile collection, how are the shape of peaks reconstructed before peak detection? Is this at all related to the profStep argument supplied to the xcmsRaw function?

Re: Quantifying centroided MS data with centWave

Reply #1 – September 21, 2014, 04:35:02 AM

XCMS quantification works with centroided data. The shape of the mass peak is not reconstructed.
The profStep parameter is not related to continuous data. It is used to create extracted ion chromatograms (mass traces, m/z slices or profiles). The quantification is based on these profiles during fillPeaks. In my experience you get better result if you change the defaults (profStep = 0.1) in xcmsSet by setting profparam in xcmsSet. For example:

Code: [Select]

xcmsSet(files=files,
              ...,
                profparam = list(step=0.005)
)

As far as I have been able to figure it is perfectly safe to set profStep very low since slices are combined to fit the ppm parameter anyway. Smaller slices uses more memory though.