Extract MS2 spectra for chromatographic peaks and features

[jorainer]

Development is going on in the ms2_spectra branch.

The aim is to extract MS2 spectra for an individual chromatographic peak (base function) as well as for a feature. The result should then be returned as a SpectrumList (being developed over at MSnbase, which allows to export them also as a mgf file.

[jorainer]

The functions are in place. @michaelwitting if you want to test: chromPeakSpectra extracts MS2 spectra for chromatographic peaks, featureSpectra for features. Both return (at the moment) a list of Spectrum2 objects (as a list for each peak/feature). This might change if I can add the SpectrumList object to MSnbase.

Also, I need to implement the custom MGF export.

[michaelwitting]

Great. I will test it later this day. Do you think that SpectrumList fits to MSnbase? For me it would fit better to XCMS, since it would be data structure for processed data... Just as a comment.

[jorainer]

I thought it would fit because it is a very general object (essentually just a list of Spectrum objects with arbitrary metadata information. Let's see - eventually I'll add to xcms.

[jorainer]

Everything is in place in the ms2_spectra branch (i.e. new Spectra object etc). Just waiting whether the class will not be included into MSnbase (could also be reused then in CompoundDb).

[michaelwitting]

The function featureSpectra throws the following error:
Error in DataFrame(feature_id = fids, peak_id = vapply(strsplit(names(res), : different row counts implied by arguments

chromPeakSpectra works fine. The writeMgfData function writes a lot of masses with zero intensities.

[jorainer]

@michaelwitting please update MSnbase (from github) and xcms from the github master branch. MSnbase 2.7.8 has Spectra present and the writeMgfData function to export Spectra objects. I made also some changes to the functions, so it could be that the featureSpectra works now. I will anyway test it on the files you provided.

[jorainer]

I don't get an error anymore with featureSpectra with the new versions. Regarding the masses with zero intensities. Yes, indeed, there are a lot of 0 in the Spectrum2 objects. One would have to clean the spectra before writing. I made a pull request on MSnbase to add a clean,Spectra method. In the meantime 0s can be removed with:

ms2_features <- featureSpectra(processedData)
ms2_features@listData <- lapply(ms2_features, clean, all = TRUE)

Note that all = TRUE removes all 0 intensities. With all = FALSE it removes all 0 intensities except those immediately left and right of a peak.

Also be aware that there would be the possibility to just report one Spectrum2 per file/feature/peak with the method parameter of chromPeakSpectra and featureSpectra.

[michaelwitting]

Okay... I will test it later. Thanks for taking care!

[michaelwitting]

I'm getting the following error when using writeMgfData:
Error in (function (classes, fdef, mtable) : unable to find an inherited method for function ‘writeMgfData’ for signature ‘"Spectra"’