[Psidev-qc-dev] QuaMeter IDFree on Metabolights data set

[David T. <dt...@su...>]

Hi, all.

During our last call, I committed to testing the QuaMeter IDFree mode on a data set from metabolomics rather than proteomics.  Last night I downloaded MTBLS548 from the Metabolights repository (https://www.ebi.ac.uk/metabolights/MTBLS548).  It's a UPLC/Waters Xevo G2 QTOF data set, using only MS; it contains zero MS/MS scans.  The total archive is around a GB in size.  I found that the "whole study" download links at the top of "Study Files" didn't work.  I looked at the corresponding FTP server, but I saw no directory for this number.  Instead I selected all the files and then used the "Download selected files" button at the bottom.  With my new 20 Mbps fibre connection at home, the whole download arrived in minutes!

I decided to translate the raw data to mzML format, centroiding peaks with the "CantWaiT" peaklister in Proteowizard: "c:\Program Files\ProteoWizard\ProteoWizard 3.0.11738\msconvert.exe" --filter "peakPicking cwt snr=1.0 peakSpace=0.1" --filter "turbocharger minCharge=1 maxCharge=4" -z *.raw
Happily, that process ran rather quickly!

>From there, I could run QuaMeter against the mzMLs pretty easily.  I used a quameter.cfg that contains these lines:
ChromatogramMzLowerOffset = "0.05mz"
ChromatogramMzUpperOffset = "0.05mz"
Instrument = "orbi"

The command line for running QuaMeter itself looked like this: "c:\Program Files\Bumbershoot\bumbershoot-bin-windows-vc120-release-3_0_11738\quameter.exe" -MetricsType idfree --cpus 1 OutputFilepath metrics.tsv *.mzML

Happily, the code ran without any apparent errors, even though it was designed to work for files containing both MS and MS/MS scans.  I am attaching the output file: metrics.tsv.

The following columns were zeros or undefined values:

XIC-WideFrac    XIC-FWHM-Q1     XIC-FWHM-Q2     XIC-FWHM-Q3     XIC-Height-Q2   XIC-Height-Q3   XIC-Height-Q4
RT-MSMS-Q1      RT-MSMS-Q2      RT-MSMS-Q3      RT-MSMS-Q4
MS2-Count       MS2-Freq-Max    MS2-Density-Q1  MS2-Density-Q2  MS2-Density-Q3  MS2-PrecZ-1     MS2-PrecZ-2     MS2-PrecZ-3     MS2-PrecZ-4     MS2-PrecZ-5     MS2-PrecZ-more  MS2-PrecZ-likely-1      MS2-PrecZ-likely-multi

Obviously an MS2 or MSMS metric cannot be computed when none of these scans are collected.  Similarly, the precursor charge metrics are undefined when no ions were selected as precursors!  The "XIC" metrics concern the peak widths of ions selected as precursors, so we lost those as well.  At present, QuaMeter makes no effort to characterize peak width or intensity for peaks that do not serve as precursor ions.

So is any of this useful?

Yes, I think so.  I am attaching a graph out of Excel: RT-TIC.gif.  The image composites four metrics: RT-TIC-Q1 through RT-TIC-Q4.  We can sum the TIC across all the MS scans to comprise a total TIC.  Then we ask what fraction of all RT duration was required to accumulate the first quarter of total TIC, the second quarter, etc.  This image shows the variation in TIC accumulation times for each of the 10 experiments.  In this image I have sorted by the startTimeStamp in the mzML files.  The authors apparently ran a random ordering of their two cohorts (file names starting with "N" rather than "H").  The time stamps tell us that half the data were acquired in the afternoon of May 3rd while the other half were acquired in the early morning hours of May 4th.  Do you see a shift in the RT-TIC plot between the fourth and fifth file?  Would we see a cohort-associated shift?  Reordering the files is just an Excel sort away!

Thanks,
Dave
The integrity and confidentiality of this email is governed by these terms / Die integriteit en vertroulikheid van hierdie e-pos word deur die volgende bepalings gereël. http://www.sun.ac.za/emaildisclaimer