Problem with error model from mProphet Sangram  2021-02-16

Hello Team Skyline,

Thank you for this great application and wish you the best in the year ahead.

I am new to proteomics and trying it with toddler's step.

So basically I have SWATH data from sciex triple TOF of cell line knock out model. I am trying to quantitate differentially expressed proteins between wild type and knock-out cells. The steps followed were mostly from the following webinar and parameter suggestions from the following paper.

So my mProphet model looks like the fig attached. I think there is something wrong as the target and decoys they have huge overlap. What might have gone wrong ?
Secondly the Pval looks absurd (can be seen in the volcano plot attached) ??
Thirdly there were some proteins like ABL1 and MDM2 which was there in the search result (generated through PeptideShaker at 1% fdr) but was missing from spectral library ?? is there any default filtering that leads into this observation ??

Thanks and regards

Nick Shulman responded:  2021-02-16
Can you send us your Skyline document?

In Skyline you can use the menu item:
File > Share
to create a .zip file containing your Skyline document and supporting files including extracted chromatograms.

If that .zip file is less than 50MB you can attach it to this support request.
Otherwise, you can upload it here:

With your mProphet model, the only thing that I could suggest is that you should look into those feature scores which are gray and in italics. Those are features which some but not all of your peptides have and are therefore not available to be used in the peak scoring model. If you select one of those scores, and switch to the "Feature Scores" bar graph, and then hover the mouse below the "Unknown" bar on the graph, you can click on the binocular icon that appears and Skyline will give you the list of peptides that are missing that score. (The list appears in the "Find Results" window which you can get to after you cancel out of the Peak Scoring dialog).

If you remove the peptides from your document that are missing a particular score, that score will become available for model training. That might result in you being able to train a better model.
However, it looks like you already have enough scores available that I would expect your model to look a lot better than it does.

I will probably be able to figure out what is going wrong when I see your file.

-- Nick
Sangram responded:  2021-02-16

Hi Nick,

So good to hear from you.

Yes definitely I can share the .sky file and it is 424Mb after zipping. But the file sharing portal of skyline cautions us on the privacy of data, since the my data has a stake from the collaborators, I would like to share it through One Drive and provide you the link in an email.

Can we work that ??


Nick Shulman responded:  2021-04-19

Thank you for sending your files. It looked like something was going wrong with your MS2 chromatograms, where most points had an intensity close to zero, but interspersed among those points were intensities that were normal, and which traced out well-shaped gaussian peaks.

I imagine the problem was that the isolation scheme specified in Skyline did not match what was collected in the raw files, but I am not sure.

Did you ever figure out what was going wrong?
-- Nick