Skip to main content ↑ ↓

Messages

This section allows you to view all Messages made by this member. Note that you can only see Messages made in areas you currently have access to.

Messages - sneumann

121
XCMS / Re: mz sorting violation
Hi,

We've now had the problem as well, on a Waters GC/MS.
The conversion included some Xcalibur peak picking, so that could be a culprit as well.
And the violating peak has an incredibly low intensity, so low it looks like a bug.

Yours,
Steffen

Code: [Select]
[305,] 322.72345 2.577887e-08
[306,] 323.53979 1.403037e-05
[307,] 324.79047 5.404045e+02
[308,] 178.08965 2.829593e-31    <--------
[309,] 326.80518 1.564260e+03
[310,] 327.61475 3.135520e-19

That's the mzXML:

Code: [Select]
<msRun scanCount="7098" startTime="PT2.0467S" endTime="PT1450.14S">
    <parentFile fileName="file://F://XXX.raw"
                fileType="RAWData"/>
    <dataProcessing centroided="1">
      <software type="conversion" name="ProteoWizard" version="2.2.3052"/>
      <processingOperation name="Conversion to mzML"/>
      <comment>Thermo/Xcalibur peak picking</comment>
    </dataProcessing>
    <dataProcessing>
      <software type="conversion" name="ProteoWizard" version="2.2.3052"/>
      <processingOperation name="Conversion to mzML"/>
    </dataProcessing>
...
    <scan num="419"
          scanType="FULL"
          centroided="1"
          msLevel="1"
          peaksCount="438"
          polarity="+"
          retentionTime="PT87.3367S"
          lowMz="40.262493133545"
          highMz="449.795928955078"
          basePeakMz="77.974868774414"
          basePeakIntensity="45256.39453125"
          totIonCurrent="3.933262835838e05">
122
XCMS / Re: XCMS and fast polarity switching
Hi,

The current XCMS allows you to read in you mzData/mzXML/mzML
with alternating polarity, then you'd use split() to create two separate
pos/neg mode xcmsRaw's, and findPeaks() on them individually.
What's currently not supported out of the box is to create two pos/neg xcmsSet's
from a directory of files, but we can talk you through creating that
from the individual peak lists.

Yours,
Steffen

P.S.: I've heard many complaints that (even fast) polarity switching will reduce
the sensitivity of your measurements. I'd love to hear your
experience, e.g. if you compare the peaks detected in a purely
positive mode measurement vs. the extracted data from switching.
If you could give me access to some raw data (any non-secret data will do,
even "only" QA mix or whatever), I'd love to have a quick glance.
124
XCMS / Re: Centwave optimal parameters
Hi Uwe,

these "insertion problems" occur if you have two masses very close (defined implicitly through the ppm parameter) within
a single mass spectrum. Usually with centroided data, that should not happen (by definition...).

In Orbitrap data that might happen due to the Orbitrap typical symmetrical artifacts (shoulders)
around high abundant peaks, and possibly the centroiding algorithm used by Thermo. If there is only a few hundred, that is not a serious problem,
and you could reduce ppm to, say, 10-15 ppm. Tens of thousands would hint at a profile spectrum.

Yours,
Steffen
125
XCMS / Re: TIC of MS/MS spectra?
Hi meow,

The split(xcmsRaw, factor) function is intended to split the "normal" MS1 scans.
How should it know which MSn goes to which of the split'ed new xcmsRaws ?
(Actually, it could possibly do so if the parent scan is properly annotated).
But it is not intended to split those MSn scans into several xcmsRaw's .

Instead, I'd suggest to do some sapply(), getMsnScan() and sum().
I have just started to fix getMsnScan(), in xcms 1.31.1 and will keep you posted.

Yours,
Steffen
126
XCMS / Re: Difference of minfrac and minsamp in group function
Hi,

both minsamp and minfrac lead to a number of minimum samples where a peak
has to be present to become a group. minsamp has an advantage if you have
sampleclasses with different numbers of samples, and you want to "fix" the
number of samples required. The relevant code reads:

Code: [Select]
gcount >= classnum*minfrac & gcount >= minsamp

so both conditions are combined via logical AND.

Yours,
Steffen
128
XCMS / Re: MRM data?
Hi nvinayav,

If you (or someone near you) has some deeper experience with R,
we could talk you through implementing this. I would need to look
how MRM is represented in these mzData (or mzML) files,
and then massage that into something that "looks" like MS1
so that the peak pickers can have a go at it. We have some experience
getting UV data into XCMS, so it is not heavy lifting, but still
not for an inexperienced R user. Are you interested ?

Yours,
Steffen
129
XCMS / Re: centWave & ROIs
Hi,

I haven't checked (just beyond dinner over here)
but you should be able to 1) say pdf(filename)
and dev.off() afterwards to save the plots to pdf.
2) you should be able to par(mfrow=c(10,5)) to place
5 by 10 plots onto one page.

Yours,
Steffen
130
XCMS Online / Re: Erro - "No MS1 data?"
Hi,

what kind of MS experiments is that ? MRM like ? Or really plain MS1 full scans ?
Can you check the mzXML file, and find those <spectrum ...>  lines,
and see whether it says something about ms level ? In doubt just post
the area at the <spectrum>.

Steffen
131
XCMS / Re: xcms documentation
Hi,

R has some extensive documentation in form of manual pages,
they take a bit to get used to, but are well structured.
Check e.g. help("findPeaks.matchedFilter") or help("findPeaks.centWave").

Also, check the papers for more on the algorithm:

  Smith, C.A. and Want, E.J. and O'Maille, G. and Abagyan,R. and
  Siuzdak, G.: XCMS: Processing mass spectrometry data for metabolite
  profiling using nonlinear peak alignment, matching and
  identification, Analytical Chemistry, 78:779-787 (2006)

  Ralf Tautenhahn, Christoph Boettcher, Steffen Neumann: Highly
  sensitive feature detection for high resolution LC/MS BMC
  Bioinformatics, 9:504 (2008)

Hope that helped,
Yours,
Steffen
133
XCMS / Re: XCMS mzML support
Hi Stephan,

Quote from: "Stephan"
The parser works fine but it takes a very long time (more than 10 min) to read in a simple 20 MB mzML file.
Is there any particular reason for that?

I can reproduce that with your file, even just
Code: [Select]
mz <- openMSfile("20101004.mzML")
takes ages. So that's a problem with either pwiz or the RAMPadapter in there,
and it needs to be fixed before XCMS could possibly adopt mzR I/O.

Thanks for your report,
yours,
Steffen
134
XCMS / Re: XCMS mzML support
Hi,

Quote from: "sneumann"
I'll try to figure how to add the packages as downloads from github,
once the packages are halfway stable. I am currently working on the windows builds.

It seems it is working on windows, so here are some preliminary downloads:

https://github.com/sneumann/mzR/zipball/0.5.4 (source package)
https://github.com/downloads/sneumann/mzR/mzR_0.5.4.zip (Win binary)

Again, you'd need my custom XCMS from github to use these.

Yours,
Steffen
135
XCMS / Re: XCMS mzML support
Hi,

Quote from: "Stephan"
Problems only start to occur when I try to get mzML support to work by out-commenting the corresponding entries in 'Makevars'.

Ah sorry, you don't do that! mzR is shipping all that pwiz stuff,
and my patched xcms is calling the IO directly from mzR.
Just install my mzR and XCMS as is, and you should be fine.

I'll try to figure how to add the packages as downloads from github,
once the packages are halfway stable. I am currently working on the windows builds.

Yours,
Steffen