How would a larger peak width vs smaller peak width affect m/z of features detected?

[Pembs]

I am trying two different peak width settings (3, 50) and (20, 60). I get more features with the 20,60. When I look at the commonality of features based on m/z (using intersect and rounding both m/z of the feature list to 4dps), I only get 10%.

Why might this be? Any points for consideration or thoughts would be appreciated please.

param = CentWaveParam(
ppm = 5,
peakwidth = c(3, 50), ##(20,60)
snthresh = 10,
noise = 10000,
prefilter = c(3, 100),
mzdiff = -0.0065,
mzCenterFun = "wMean",
integrate = 1,
fitgauss = FALSE)

[jorainer]

I would suggest to set the peakwidth setting to half of the observed peak width of some compounds (e.g. internal standards or other compounds you might see in the data) up to two times the observed peak width. From own experience: if the difference between the min and max peakwidth is very large, centWave tends to miss peaks (has to do with the internal continuous wavelet transformation - an ~ fixed number of scales is used, end they might be a little too crude (far spaced) for a large difference of min to max peakwidth).

why you see so large differences I don't know. did you only consider m/z or also the rt of the detected peaks?

Also, the noise level you set is pretty high, any reasons for this? I would rather change the prefilter instead (which has then an influence on the run-time, because fewer ROIs will be defined in which peak detection is performed).

Also your ppm is pretty low - do you have orbitrap data? for TOF I would suggest to increase. I usually use ppm = 40.