Using IM predictor on DIA PASEF

Using IM predictor on DIA PASEF paga  2021-12-30

Hi again Skyline team, I come with a new inquiry.

I've been going over the 'Skyline Ion Mobility Spectrum Filtering' tutorial to try and implement the IM dimension into my DIA-PASEF experiment, hoping it could lower the noise and therefore help on the identification of our 30 Heavy labeled peptides and their non-labeled counterparts. I realize that the tutorial I'm referring to is not directly showing this kind of application, but in theory, it should work as well.

So far, my attempts have only worsened identification when compared to a spectral library-based predictor, which can be seen in the picture attached, where I totally lose the signal on an otherwise confidently-identified peptide. Based on this, I come with the following inquiry. What is the nature of the Q_2014_0522_ASMS2014 library? Is this a file acquired in DDA acquisition mode, while BSA_Frag_100nM_18May15_Fir_15-04-02.d and Yeast_0pt1ug_BSA_100nM_18May15_Fir_15-04-01.d files are samples acquired in DIA mode?

If this is the case, then the difficulties I'm experiencing may be because the DIA file I'm using to train the IM prediction is not a pure sample (it contains Hela extract) and therefore selects the wrong drift time range.

Once again, thanks for your help.

Brendan MacLean responded:  2021-12-30

Hi Pablo,
That tutorial is intended for more traditional targeted approaches where you can create a purified sample, possibly with labeled reference standards or light synthetic peptides for your targets of interest.

For your application, you will be much better off referring to the tutorial and webinar on diaPASEF:

Hope these help you achieve what you are hoping for. Thanks for your interest in using Skyline in your research.


paga responded:  2022-01-03

Hi Brendan,

I have actually gone in-depth in both the webinar and the tutorial. Fantastic work by the way! The issue is that on the webinar, you only touch on IM right before Q&A, so you basically only show how to find the graphs Ben talked about in the introduction. In these graphs, I can already see there is some IM filtering, but it's not clear how that filter was applied.

It's the same issue when going over the written tutorial (pdf). At any point during the Import Wizard set up I see any settings regarding IM, and yet on page 25 when the IM 3D plot is shown, a restricted IM range is already being applied. Is this maybe something that's done automatically while using the Import Wizard? Normally I work without it and load a spectral library through Settings --> Peptide Settings before importing Results, since this doesn't require for me to upload the library each time I want to run an analysis.

Overall, is the tutorial missing a step on how to set up the IM filtering, or does Skyline do this automatically through IW?

Brian Pratt responded:  2022-01-03

Hi Pablo,

I'm wrapping up some work on improving ion mobility handling in transition list and assay library imports, I'd be very interested to see an example of the files you're bringing into Skyline to make sure that any ion mobility information they contain is retained and properly used.

This Skyline Tip might be helpful as well: Mobility Separation (IMS%2C or Drift Time) Data

Best Regards,

Brian Pratt

paga responded:  2022-01-03

I have uploaded a zip file though your platform with the DDA file and a DIA run, together with a list of the 30 targets spiked Heavy Labeled peptides. It's called paga_zip. Both have iRTs from Biognosys (C11).

Brian Pratt responded:  2022-01-03

Thanks for the files.

The issue here is that there is no provided information about the expected ion mobility or Collision Cross Section of the targets, so Skyline has no way to take advantage of the ion mobility dimension in the raw data. Do you perhaps have those values anywhere?

paga responded:  2022-01-03

Unfortunatelly no.

The colleague that got the results is out on holiday, and will be for a while. Will get back to you when I have them, thanks for the help!

Brian Pratt responded:  2022-01-03

Returning to the top of this thread (I came in late) I see that you have already experimented with training an ion mobility library on this data. Could you perhaps provide the Skyline document (using File > Share > Complete) that includes that ion mobility library? Perhaps there's a setting that can be tweaked for a better result.

paga responded:  2022-01-04

I have uploaded a Skyline run where I did the IM training with the 3 replicates of the DIA file uploaded before, and obtain a much worse heavy peptide identification than the spectral-library predictor. The name of this new file is paga_IM_zip.

Brian Pratt responded:  2022-01-04

Thanks, I'll have a look at this today.

I wanted to mention, though, that spectral libraries and ion mobility filtering are not mutually exclusive - they're actually complimentary.

  • Brian
Brian Pratt responded:  2022-01-04

I don't see that file anywhere - can you try uploading again? Thanks.

paga responded:  2022-01-05

I didnt know they weren mutually exclusive, thanks for that. Maybe thats the issue then. The file is now uploaded.

Brian Pratt responded:  2022-01-05

Can you send that again, using Skyline's File > Share > Complete menu item to produce a file? That eliminates the chance of missing files, as in the current upload (missing ion mobility library).



paga responded:  2022-01-11

Apologies for the delay

I realized the lack of library was my bad, it should be there now that I reuploaded it again under the name paga_IM_zip_2

Brian Pratt responded:  2022-01-11

Hmm, that's not the ion mobility library used by that Skyline document. Did you actually use Skyline's File > Share > Complete menu item to create the zip file, as requested?

Also I don't have the raw data used by this Skyline document (that's normal, raw data files are not included in files, but neither are they the data you previously provided):

paga responded:  2022-01-12

Ok last try,

I follow the zip file creation advice through File > Share and put it together with the skyline project, DIA raw files libs and IM libraries in a single zip file I have uploaded through the websites file-sharing options under the name paga_IM_allfiles.

Brian Pratt responded:  2022-01-12

OK, the file seems good (there's no need to separately include .sky, .blib, .imsdb etc, they're all in the generated but those .d files seem to be corrupted. Can you verify them then resend?