• Sign In
  1. Requests

Peptide from PEAKS exported data not showing up in Skyline library

support

View Request

view list print
Peptide from PEAKS exported data not showing up in Skyline library andrea-patterson  2015-07-01 11:46
 
Hi,

My first attachment is a screenshot of the PEAKS peptides immediately before exporting them for Skyline. There are 15 peptides here, and note that it includes the highlighted peptide "KIFGSLAFL" (unmodified, and only 1 K at the beginning).

The second attachment is the Skyline library built from this PEAKS export. It shows 15 peptides, but note that one of them (the double-K version of the above peptide) has two charge states shown, and that the single-K, unmodified peptide KIFGSLAFL is not there.

When building the library, I have the cutoff score set to 0 so that I should get every peptide from PEAKS. Is there some reason you can think of why this one peptide is not transitioning into my Skyline library? (I need it!)

Thank you :)
Andrea
 
 
Kaipo Tamura responded:  2015-07-01 12:00
Hi Andrea,
Missing peptides from a library usually means that multiple peptides matched the same spectrum, but it might not be the case here. Can you email me the pepXML file to kaipot@uw.edu?

Thanks,
Kaipo
 
Kaipo Tamura responded:  2015-07-01 16:29
Hi Andrea,

I've had a look at this and it turns out that the peptide is actually in the library (i.e. the .blib file), but Skyline has issues dealing with spectra in the library that contain more than 65,535 peaks. This particular peptide has 117,278 peaks for the selected spectrum.

I think this can be fixed in the code pretty easily, but in the meantime you can work around the issue by removing PSMs from the pep.xml file that correspond to problematic spectra so that a different spectrum is selected for the peptide. For example (regarding KIFGSLAFL), if spectrum 84205 (which has 25,244 peaks) gets selected instead of 82641 (117,278 peaks) the issue will go away. Sorry for the inconvenience - I can share a library like this with you if you'd like.

Thanks,
Kaipo
 
Brendan MacLean responded:  2015-07-01 21:20
Hi Andrea,
I will note that spectra with so many peaks are unlikely to work well in Skyline for the ways that it will use them. You will be much better off centroiding your spectra. You should be able to do this before you search them in PEAKS. Is there any reason that giving PEAKS profile spectra would improve the search? If so, then you may want to ask the PEAKS team for a way to export centroided spectra for Skyline. Otherwise, just go ahead and centroid your spectra before you search.

We will certainly fix this confusing issue, but it that is likely to just give you new insight into why profile library spectra do not work well for Skyline.

Thanks for your clear report and for providing your PEAKS data files.

--Brendan
 
andrea-patterson responded:  2015-07-04 09:22
Thanks Kaipo and Brendan,

I have never altered a pep.xml file before, but I just opened it in notepad and deleted all the instances of KIFGSLAFL other than the one you mentioned above (spectrum 84205). I was guessing at the exact number of lines to delete for each, but it worked! I didn't see a line that indicated the number of peaks included in each spectrum, and I am guessing there is a program I could download to view this type of information and manipulate pep.xml files more easily? What would you recommend (for next time)?

I will have to look into the centroiding - I don't know if that would interfere with searching in PEAKS or not. That is great to know though, thank you for the helpful insight!

Andrea
 
Kaipo Tamura responded:  2015-07-06 10:55
Hi Andrea,
The pep.xml doesn't look like it contains the number of peaks in the spectra, though the information is in the spectrum file. A program like SeeMS (included as part of the ProteoWizard project: http://proteowizard.sourceforge.net/) can open mzXML files and provide this information - it has a column for the number of peaks in each spectrum. Hopefully that is somewhat helpful.

Thanks,
Kaipo