Export of non-existing ion mobility data

[michaelwitting]

I'm using the MSnbase writeMSData function to export to .mzML after some processing. Using this export the .mzML than contains information on ion mobility, thought it was only normal QToF data. Values are set to nan.

Here is an example: <cvParam cvRef="MS" accession="MS:1002476" name="ion mobility drift time" value="nan" unitCvRef="UO" unitAccession="UO:0000028" unitName="millisecond"/>

Discussion with @jorainer have led to this issue.

Any ideas?

[michaelwitting]

PS: Proteowizard and MSConvert are version 3.0.21079

[jorainer]

what version of mzR are you using?

[michaelwitting]

2.28.0

[sneumann]

Can I have some context ? Would that happen if you exported faahKO that way ? Or was there ion mobility in the original mzML, removed/dropped while processing, and now turning up as NaN ? Yours, Steffen

[jorainer]

Reproducible example:

library(MSnbase)
fl <- system.file("sciex", "20171016_POOL_POS_1_105-134.mzML", package = "msdata")
data <- readMSData(fl, mode = "onDisk")
data <- pickPeaks(smooth(data))
writeMSData(data, file = "test.mzML", copy = TRUE)

the mzML file contains now this funny ion mobility drift time spectra variable.

[michaelwitting]

No, there was no ion mobility data in the original files. It is from a Sciex X500R QToF. I perform smoothing and centroiding using MSnbase and then export to .mzML. I wanted to import a testfile into mzMine for some visualization then got the complain of the software that it cannot deal with nan

[jorainer]

note that the ion mobility drift time is a default header column (spectra variable) that is returned by mzR::header (similar to others like "scan window lower limit", "scan window upper limit" etc). In theory having this header info in the exported mzML file should not hurt - since it is supported by mzML.

[michaelwitting]

I guess it is normally no hurting, but mzMine for example cannot deal with the nan as value in there.

[jorainer]

@sneumann , do you know by chance what the correct encoding for a missing value would be in mzML? I assumed that proteowizard would handle that correctly...

[jorainer]

Hm - strange. If the ionMobilityDriftTime is NA it should actually not be exported: https://github.com/sneumann/mzR/blob/master/src/RcppPwiz.cpp#L768:L770 - there must be something strange going on.

[michaelwitting]

I checked the original .mzML file, there is no single mobility thing mentioned... strange... I have it already after reading the data with readMSdata, but all the entries are NA.

[michaelwitting]

After performing smoothing and centroiding, still all of the are NA

[michaelwitting]

I checked it via @featureData@data

[jorainer]

yes, it seems to be a bug in my Rcpp code - in fact, NA variables should not be exported, but it seems I'm not correctly testing for NA.

[jorainer]

Fixed in PR #267