Skip to main content ↑ ↓
Topic: centroid vs profile: What parameters? (Read 11886 times) previous topic - next topic

centroid vs profile: What parameters?

I am a new user to xcms online.  I converted profile MassLynx data to mxXML.  I noted that the canned parameters seem to use centroid data via centWave. First naive question:  Is product of conversion to mxXML profile or centroid?  Next, should I be using centWave or matchedFilter?  If matchedFilter, what is a good parameter set for HPLC ms only proteomics data?  Other suggestions?

Thanks

Re: centroid vs profile: What parameters?

Reply #1 –
mzXML can contain either profile or centroid mode data.
The vendor raw data often contains both (check settings) or centroid spectra are calculated during conversion.

XCMS methods can be applied for proteomics data, but it is not optimized to deconvolve the isotopic pattern of peptides, i.e. you will see each isotopic peak as a separate feature.
It will try to annotate isotopic peaks (M, M+1, ...) but XCMS/CAMERA does use any peptide specific model for that.

In general, I always recommend using centroid mode data in combination with centWave feature detection which has shown to be most sensitive.
Also, files will be much smaller in centroid mode, i.e. upload much faster.

Re: centroid vs profile: What parameters?

Reply #2 –
Thanks Ralph:

Uploaded data was profile.  I understand there that xcms does not deconvolute and am not worried.  But I would appreciate your suggestions on parameters for matchedFilter data for HPLC Qtof data.

I do plan in near future to try centroid, but prefer to continue with profile to meet internal needs.

Thanks again.

Re: centroid vs profile: What parameters?

Reply #3 –
Hi Will,

I am going to create a separate parameter set for high resolution data in profile mode using matchedFilter.

However, using centWave will provide additional features like the spectrum view in the results table.

If you convert using proteoWizard as described here https://xcmsonline.scripps.edu/docs/fileformats.html 
centroidization can be selected as an option during conversion. 
If you convert using the instrument software there should also be an option or checkbox (often also called "peak data" vs full spectrum or profile mode.) for that.

Ralf

Re: centroid vs profile: What parameters?

Reply #4 –
Hi Ralf:

Thanks for info about converting to centroid from profile.  Will try.  However, please note that ProteoWizard corrupted data in first use and you suggested to convert to mzXML from MassLynx, which worked.

Also, would greatly appreciate your effort to improve processing of profile data.

Re: centroid vs profile: What parameters?

Reply #5 –
So if we can deisotope our data (ie remove isotopic peaks from the peptides) prior to processing by XCMS (and we also centroid the data), would XCMS be "perfectly" capable of doing the alignment and feature detection for LC-MS based proteomics data as well as it does for metabolomics data? That is, is there any special algorithm that one knows of that "proteomics specific LC-MS alignment programs" would use?

Re: centroid vs profile: What parameters?

Reply #6 –
Please do not hijack old threads. Create a new thread in the appropriate forum for your question.