MSFraggerGlyco to Skyline document

[juank1892]

- Upload your log file
(If a log file hasn't been generated, go to the 'Run' tab in FragPipe, click 'Export Log', zip the resulting "log_[date_time].txt" file to avoid truncation, then attach the zipped file by drag & drop here.)
log_2024-08-29_17-44-58.txt

- Describe the issue or question:
Hi there,

(Beforehand apologies! I have not played with FragPipe in quite a while!)

I downloaded the dataset of this publication https://pubs.acs.org/doi/10.1021/acs.analchem.3c05874 with PXD047898. I was able to re-run the search of some of the files and could identify some glycopeptides although not comparable number to those reported. I also tried using the new Skyline document creation option, but noticed that no glycopeptides made it to the document.

I have attached the log file to see if you can point me out to some settings that are not optimal.

When creating a spectral library for Skyline, which score should be used to filter the output?

Please let me know what other information would you need me to share to solve this concerns.

Looking forward to your response.
JC

[dpolasky]

Hi JC,

DDA glyco searches currently require a less than ideal workaround to get the glycans into Skyline - you can find the details here: https://fragpipe.nesvilab.org/docs/tutorial_glyco.html#examine-the-results. We're developing an improved workflow now but it's not quite ready yet.

Regarding the differences vs the published result, I believe in the paper they used FragPipe as well so it should be possible to reproduce their results. If the PXD repository provides their FragPipe workflow file, you can download it and load it to search with the exact settings they used. If not, I would try to match the settings to what they describe in the paper - it looks like they used a different glycan database and changed some variable mods and other settings vs our defaults.

Best,
Dan

[juank1892]

Hi Dan,

Thank you for pointing out the workaround. Now I am able to get the identified glycopeptides into the Skyline file written by FragPipe.

It is nice that it can process raw timsTOF files, but I would like to inquire a few things.

• Usage of ion mobility library: the spectral library has the 1/K0 value assigned for each precursor, but it is not being used when generating the EICs as an ion mobility library is not being created. Was something else required to be enabled to make use of this or generation of a Skyline .imsdb under development?

• Score threshold for spectral library: how is the score threshold decided for the created spectral library? I see that a PeptideProphet confidence score of 0.4303 was used for my current output, but that seems quite a lenient score, no? However, most non-glyco peptide in the document have a score >0.95. But the glycopeptides had lower PeptideProphet scores which only show up in the library with such low PeptideProphet score.

• Peak boundaries: the peak boundaries created for Skyline are those assigned from the Skyline peak picking algorithms or are the ones defined by the quant modules of FragPipe?

• You are nicely using oxonium ions for scoring and visualization in FragpipeViewer. However, these are not included in the produced skyline documents. Attached you will find a template that has these oxonium defined under Settings -> Transition Settings -> Filter -> Special Ions

Maybe you can edit this in the Skyline xml file that you are creating based on the ones in FrapPipe input.
GlycoProteomics_template.sky.zip

If this conversation is getting too long, let me know if you would be willing to set up a Teams/Zoom meeting to clarify questions faster.

Thank you for your time and help. Looking forward to exploiting FragPipe+Skyline potential in glycoproteomics!
JC

[dpolasky]

Hi JC,

Thanks for all the feedback! Some of these are just a result of the workaround, and will be fixed once the final workflow is ready. E.g., the ion mobility can be read in from the spectral library for non-glyco workflows, but isn't included because of the workaround for glyco. The score threshold is determined automatically from the FDR filtering, and corresponds to your set FDR level (default is 1%). However, because of the workaround, glycan composition level filtering is not applied and there are (slight) differences in how the scores are calculated for peptides because the extended mass model isn't used in PeptideProphet.

The peak boundaries in this case are determined by Skyline. For DIA data, the peak bounds from DIA-NN can be imported to the document, but we have not yet implemented a way to import the DDA peak boundaries from IonQuant.

We can certainly add oxonium ions to the Skyline documents - will work on that for the next version.

If you're interested in testing a beta version with these new features, I can let you know when we have it ready and you can give it a try - just let me know.

Best,
Dan

[juank1892]

Hi Dan,

Sure thing! I would be happy to test the beta version when the new features are ready!

I already see FragPipe as a nice addition to our workflows and would be happy to smooth its communication with Skyline where we ultimately do all of our data validation and integration.

Feel free to reach out to my work e-mail if you need a more direct communication: [email protected]

JC