Skyline for peptide search using Waters MSe data

Skyline for peptide search using Waters MSe data ywu1  2021-06-07

Hi there,

Curious to the ability of Skyline to perform a peptide search using MSe data generated on a Waters Synapt G2Si. Reading some of the older posts, it sounds like PLGS was the preferred route prior to Skyline, just curious if this was addressed since you have added so many features as of late. Additionally, If this is indeed possible, would skyline be able to read the ion mobility data as well?


matt.chambers42 responded:  2021-06-07

The DDA search added to Skyline recently won't work with MS^e. The DIA-Umpire support added even more recently only supports SWATH and V_SWATH modes. Although the standalone DIA-Umpire does support MS^e, I've never tested it so I don't know how it would compare to PLGS.

I'll ask Brendan how to prioritize adding support for MS^e but for now I'd say PLGS is still the preferred route.

Juan C. Rojas E. responded:  2021-06-08

Hi Christopher and Matt,

If you are aiming for maximum coverage I would not advise using DIA-Umpire since it cannot support IMS information and the best results in identification (and quantitation if you are aiming for low abundant peptides) are obtained when you perdorm IMS-MSE (a.k.a HDMSE) experiments. I tried DIA-Umpire with some MSE files and the results were poor to say the least, but it could be I didn't find the optimal settings and didn't invest enough time into it.

That I am aware of, only Apex3D can support the data deconvolution of Waters data containing IMS information. Apex3D is a subprocess that is run prior to searches with PLGS to generate the pseudoMSMS spectra of features alligned in the RT and ion mobility dimensions.

From the PLGS software you can export the .mgf files generated by Apex3D to perform searches in other search engines if you prefer. Personally, I prefer searching the .mgf files in other search engines because the output of PLGS ("*_final_fragment.csv" files) only shows the matched fragments, making you essentially blind to all the other signals present in the allignment that might be important in confidence assignment of the peptide. The problem with the .mgf output of PLGS is that it performs charge deconvolution both for the precursor and fragment ions (I assume to facilitate identification) which would prevent you from mapping back the identifications back to the raw files to generate the EICs.

Alternatively, you can run Apex3D from the command line to generate the .mgf files as input (e.g. for MSAmanda on Skyiline) without applying charge deconvolution to the precursor ions. However, we haven't found a way to not apply charge deconvolution to the fragment ions so any multiply charged fragments that have been deconvoluted would be missed.

Finally, Apex3D also generates a "*_Pep3D_Spectrum.xml" output that contains all the binned information necessary (i.e. alligned ions dected in the low energy, high energy scans, and same IMS bin (if used)) to generate the pseudoMSMS spectra. However, a parser would be required to extract the information.

By the way, Skyline supports Ion mobility data from HDMSE files generate from Synapt G2-Si instruments! It is beautiful to finally be able to visualize the real data instead of relying on the PLGS and other Waters supported blackbox softwares.

Hope this information helps somehow.

Matt, if this topic moves forwards I would be more than happy to provide access to data related to these topics that we have collected and curated if it is needed.


Brendan MacLean responded:  2021-06-09

Yes, definitely start with PLGS for your peptide search. After that, you can treat the search results as if you are building a spectral library on your results files, as has been common with DDA for MS1 filtering for a long time, with PRM for MS/MS also, and with DIA-Umpire for DIA most recently.

This ASMS 2012 oral presentation attempts to cover these concepts for early DIA and standard MS^e, where I essentially predicted the coming of tools like DIA-Umpire for SWATH-type DIA.

Yes, the same approach now works with HDMS^e, where you can consider PLGS either the inspiration for DIA-Umpire, or it equivalent in an HDMS^e targeting workflow in Skyline.

Hope between Juan and my answers you have enough to continue your assessment of Skyline for your application. Thanks for posting to the Skyline support board.