Re: Cross MS-Platform Quantitation with XCMS, HOW?
Reply #2 –
Your image did not make it into the posting for some reason.
Both feature detection algorithms, matchedFilter and centWave, use the same type of normalization
pwid <- (scantime[peakrange[2]] - scantime[peakrange[1]])/(peakrange[2] - peakrange[1])
peaks[p,"into"] <- pwid * sum(d[lm[1]:lm[2]])
to compensate for scan rate changes.
Colin came up with this normalization factor and I used the same normalization for centWave,
to make the intensities calculated by matchedFilter and centWave comparable.
I think the initial idea was to compensate for inter-sample scan rate changes,
e.g. running th same sample with the same gradient with e.g. 1.5 Hz or 2 Hz scan rate should result in similar
feature intensities, although the number of centroids that are integrated for each feature will be different.
It was designed for Q or TOF data, not for Orbitrap.
I am not sure what "a connect the centroids model" is supposed to mean.
Sounds like either fitting a model or just summing up the intensity values of the centroids.
The latter can easily be achieved by just setting pwid to 1 in the source code.
If a different type of normalization turns out to work better with Orbitrap data
then I would be happy to include it as an option for centWave.