Building libraries

Building libraries luzjpaulo25041  2021-11-05

Dear MacCoss lab Team,

I am writing because I am trying to build a library and the result seems unexpected. I have some results from Peaks DB of a Bruker impact II TOF search. I feel like the quality of the library is not what is supposed to be. There are few peptides and the spectra matches are odd. I would like to know if there is a a problem with the run or something else. Could you help me?

Thanks for your attention.

Best Regards,

Joao Silva

Kaipo responded:  2021-11-05

Hi Joao,
The library builder might be matching PSMs to the wrong spectra.
Would you mind uploading the pepxml and spectrum file(s) to:


luzjpaulo25041 responded:  2021-11-06

Hello Kaipo,

I uploaded the rar file named Search_joao_peaks with the requested data. If you have any problem to access it let me know.
Thanks for your help.

Best regards,

Kaipo responded:  2021-11-09

Hi Joao,
It looks like the PSMs are being matched to the correct spectra. There are many modifications on the peptides and one issue might be that in some cases multiple modifications are occurring at the same amino acid, which Skyline has trouble with (you'd need to create new modifications that are combinations of the individual ones).
There are 396 PSMs in the pep.xml file, including 298 unique peptides. When I build a spectral library, I end up with a library containing all 396 PSMs. So the count seems okay.
You are right that some of the spectra don't seem very good, but as far as I can tell they are as provided in the mzXML files so it's hard for me to say what the issue is there, if any.


Juan C. Rojas E. responded:  2021-11-10

Hi Joao,

Do you have multiple MSMS spectra for the same precursor ion across the chromatographic peak? By any chance you ticked "combined MSMS spectra" in the "Data pre-processing" step of PEAKS?

Do the PSMs look better in PEAKs studio when you untick "preprocess"?

I found an issue with PEAKS that rings a bell to what you are describing and reported it to BSI (see below description), but I never got a follow-up. My work around was to disable "Merge MSMS spectra" in the data preprocessing step.

"I just encountered an issue with the exports I get from PEAKS for Skyline. After importing the results into Skyline I noticed that there were many PSMs close to the apex that should have been possible based on the short dynamic exclusion times I used and the low complexity of the sample (HSA digest).

For example, peptide K.QNCELFEQLGEYK.F [413, 425] (Slide 1) for file S4_21_7_QC3.raw was mapped with only one MSMS spectrum of really low intensity and is missing many PSMs of considerably better quality close to the apex that are identified by PEAKS (for example Slide 2) and is actually indicated in the PSMs table (Slide 3; attached .csv table).

When I check for the MS/MS scans proposing this peptide in the .pepXML export I realize the issue might be due to how the grouped spectra are reported (Slide 4; attached .pepXML). I think Skyline only read the scan number written in "start_scan" to generate the library spectrum for this peptide. Due to its quality I would have disregarded this peptide in follow-up data validation steps.

The files were processed with "Merge Scans [DDA]" ticked and when comparing PSM outputs I noticed it really helps in controlling false positive rates.

However, for Skyline I need to be able to access all PSMs used to make sure my spectra libraries are representative of the information used by PEAKS. Is there a way of exporting ALL PSMs and not the best representative one noted in the PSM tables? "

(see attached powerpoint with example)