That is a lot of features per sample. I've not seen data from a LECO TOF before, but one of the first things I would check is the quality of the centroidization algorithm. How do the centroid spectra look like compared to the profile spectra. Sometimes artifacts (split peaks, ringing effects, etc.) are generated when centroidization does not work that well. Does it apply some kind of thresholding? If you feel like centWave is picking up too many features with very low intensity you should play with the prefilter option.
As you noticed in your other posting, the step size ("step"), i.e. the bin size of the profile matrix has an influence on the result.
Which "step" size did you use here ?
It should be small enough so that these features won't fall into the same bin.
An alternative would be to implement a "rawEIC" method for xcmsSet, i.e. not to use the profile matrix but the full raw data (rawEIC for xmsRaw) for EIC generation. I think this might be nice to have anyway, especially in combination with centWave, so I'll add this to my personal to-do list.
In which column do you see the K+ ? If it's in the adduct column then this is on purpose - XCMS Online is trying to deconvolve and detect (using CAMERA) possible adducts, including pretty much all possibilities. But for the METLIN search it will only use the additional adducts that you selected in the identification tab.
You reduced the accuracy for feature detection from 30 (in the Agilent TOF parameterset) down to only 10 ppm, which is most likely the reason that not all features were detected.
XCMS online shows you the neutral mass in the adducts column whenever possible. Sometimes this information is not available when only single peaks/feature are observed. It is mathematically not possible to calculate neutral mass from a single peak. It needs multiple peaks per feature to calculate, otherwise all you can do is to search databases using the most likely adducts like M+H and M+Na. This is what XCMS online is doing.
e.g. 3 UP 5.9 7.12731e-7 859.4871 126.49 1,362 2,432 14,330 [M+H]+ 858.482 where 859.4871 is m/z (M+H) and 858.482 is the neutral mass.
You can only get putative IDs/database hits using exact mass. I don't know what Mass Profiler is showing, but it is not possible to identify compounds based on exact mass only.
2. These are putative IDs which have to be verified using e.g. MS/MS data. Also see here 3. These are different retention time correction methods :
the
obiwarp method is based on correlations of the raw data, method is described in Prince et al. the peakgroups method uses "well behaved" peak groups and nonlinear regression to calculate retention time deviations for every time point of each sample, method described in the original XCMS publication by Smith et. al.[/list]