Hi Zhenbin,
I would recommend you simply use the approach outlined in both webinars 14 and 15 to use a native Skyline spectral library built from your search results. However, there are "assay libraries" that have been made publicly available and also a paper from the Aebersold lab detailing how they build their assay libraries. (https://www.researchgate.net/profile/Ludovic_Gillet2/publication/272187418_Building_high-quality_assay_libraries_for_targeted_analysis_of_SWATH_MS_data/links/56960cb108ae3ad8e33d9dc3/Building-high-quality-assay-libraries-for-targeted-analysis-of-SWATH-MS-data.pdf) Much of the advice in this manuscript should be generally applicable, but the paper also describes the tools they use to produce the actual tabular format imported into Skyline during webinar 14 (and for the Navarro, Nature Methods 2016 paper).

Thanks for your interest in using Skyline for querying DIA data.

--Brendan

Hi Brendan,

Thank you for the recommendation!

I am using the approach outlined in webnars 14 and 15. But I have no idea how to produce the transition list that was imported into the Skyline as shown in the attachments.
I have another question, how to build my own iRT library? I selected over 500 hundred peptides for DDA result and calculated the iRT values in the Skyline. But I don't know how to build a iRT library with them.

Thank you!

Zhenbin

There is only one case where I import a transition list in these tutorials. (one produced by a collaborator - Pedro Navarro) If you are not producing one of these transition lists (a.k.a. Assay Library) then I suggest you not try to produce one, but instead follow the workflow based on building your own spectral library in Skyline.

The approach where I show a pre-created Assay Library is really more for people who have these Assay Library's to start from using other tools or following the protocol in the paper I cited. If you really want one of these, then you should follow that protocol.

However, if you are willing to simply build a spectral library from within Skyline, then adding targets to your Skyline document becomes the same as outlined in the MS1 Filtering tutorial or Targeted Method Editing tutorial.

--Brendan

Hi Brendan,

Thank you for the reply!

In the tutorial, said "You can even use a set of peptides endogenous to a particular experiment as your
standard, as long as they can be consistently measured and span most of the gradient range you are
attempting to predict."
I am wondering how to build my own iRT library without iRT kit standards?

Thank you!

Zhenbin

Hi Zhenbin,
This is a good question and one that seems to be coming a lot more lately. Perhaps a good topic for a new webinar. At this point, your best resources for this would be to consult the iRT tutorial:

http://skyline.ms/tutorial_irt.url

And the existing iRT webinar:

http://skyline.ms/webinar07.url

The example in these two resources does use an iRT standard mix, but they show how to calibrate an iRT library from scratch based on a set of peptides.

The only missing piece for using endogenous peptides is coming up with a set of peptides that meat the criteria:

1. Consistently measured
2. Well spread across your gradient

If you follow the iRT tutorial and you click the Calibrate button when you have a lot of peptides in your document, Skyline will offer to choose some that span the gradient. That means you mostly just have to create a document filled with "consistently measured" peptides. You can do this by searching your entire library in some of your samples using full-gradient extraction (no RT prediction) and then reducing the document to only the peptides with very high scores, ideally in multiple runs. Once you have your set of "standards", you can then use the "Add..." button and choose "From Library" to produce iRT values for the rest of the peptides in your library.

You may then want to test your standards in some other replicate runs and make sure they all get detected consistently. I did this recently. I started with 100 peptides and then reduced the number to 72. Once you have an iRT library, it is relatively easy to swich which peptides are the standards by clicking a button in the Edit iRT Calculator form and pasting a list of the peptides you want to be standards.

We will try to make this easier in the future, as it seems you are not the only one interested in it. Thanks for posting your question to the support board.

--Brendan

Hi Brendan,

Thank you! I have two more questions:
Could I search my entire library in some of my samples using full-gradient extraction in Skyline?
Could I use as many standards as possible to make sure some of them could be consistently measured in my samples? Maybe it doesn't work. I tried but failed, as shown in the attached figure.

Best regards,

Zhenbin

1. Yes. I believe this is how I started myself recently. I even ran mProphet on data imported this way, reduced my document to only several hundred peptides with the highest Detection Z Scores, and then used those to get 100 peptides across the gradient (using the Calibrate button in the Edit iRT Calculator form). I then further reduced the set to 70 that were consistently identified in 3 replicate runs.

2. Just including a lot doesn't work with the iRT method, if they are not consistently detected. The Skyline algorithm does try to throw out outliers, but things work far better if you just have highly consistently detectable peptides and don't ask Skyline to guess the linear trend from peaks which may be only noise. And as you have seen, if the chosen peaks cannot produce a 0.99 correlation with 80% of your standard peptides, Skyline just gives up and fails. That is what Skyline is telling you with "r = 0.973 < 0.990 (at 296 points minimum)" in the image you provided. You probably want to at least understand why so many points have no measured time at all, indicating Skyline never saw anything for these. Perhaps you are including precursors outside your measured range? But, also, you should consider wether the linear trend Skyline is choosing after removing outliers really reflects the gradient. Certainly, it would have an easier time with more consistently detectable points.

Good luck refining your list of endogenous standard peptides for use with iRT.

--Brendan