That particular line item would not have an impact on anything that you are doing.
That item was basically a bug fix about the scenario of training a "Default" peak scoring model at "Refine > Reintegrate" in order to generate q-values. It was mostly an academic change: if you were actually interested in improving your peak picking, you would be training a "mProphet" peak scoring model, which was unaffected by that change. If you want to learn more about training peak scoring models, you can look at the Advanced Peak Picking models tutorial:
https://skyline.ms/wiki/home/software/Skyline/page.view?name=tutorial_peak_picking

I cannot think of any improvements in Skyline's automated peak picking that happened between Skyline 21.1 and the current version which is 23.1.
One thing that we added in Skyline 21.2 is a "Synchronized Integration" menu item which can make manually curating your peaks easier.

Also, in Skyline 22.2, we added the "Candidate Peaks" grid (View > Other Grids > Candidate Peaks). This window can give you a lot of insight into why Skyline has chosen the particular peak that it did.

We have plans to improve automated peak picking in the next version of Skyline that we are working on now. In particular, this will involve using information from the replicates where the peptide was easy to find in order to decide which peak to choose in the replicates where the peak was not detected.
-- Nick

Thanks for confirming no impact. I was not aware of the Synchronization Integration menu item. I have used 'apply all' in the integration menu to quickly align peak integration across replicates to one good replicate. How does synchronization differ to this?, i see the alignment options are 'None' and 'RT calculator', do you need to setup RT calculator... Is this covered in the tut you shared, or is there a short blurb on it somewhere, thanks, appreciate feedback

The big difference between "Apply peak to all" and "Synchronized Integration" is that once you have synchronized integration turned on, it happens ever time you change peak boundaries, and immediately applies the results to all of the other replicates, as opposed to "Apply peak to all" which only happens when you click that menu item.

The other thing is that "Apply peak to all" only chooses a peak in the other replicates from the list of already detected candidate peaks. With "synchronized integration", Skyline just chooses the exact same start and end times in each of the other replicates.

Synchronized integration works a little better if you align to an iRT calculator, because, then, instead of choosing the exact same retention times in each replicate, Skyline will choose the times that coincide to the exact same iRT values in each replicates. This works better if there is a bit of retention time drift between the replicates that can be compensated for by your iRT standards.
-- Nick

Was your original question asking whether Skyline 22.2 will produce the same numbers as Skyline 21.1?
The answer is "probably not" but it is difficult to say exactly how the numbers will be different.

I cannot think of anything specific that might have changed, but I think there have been some bug fixes that would only affect rare edge cases. The thing is, whenever you have a large dataset, the probability of having a rare edge case increases a lot.
My guess would be that any differences that you see in your numbers between Skyline 22.2 and 21.1 will involve cases where the peptide probably wasn't even present, and Skyline 21.1 picked one very bad peak, and Skyline 22.2 picked a different equally bad peak.

If you ever want to know why Skyline two different versions of Skyline are getting different numbers in a particular case, you can always send us your Skyline documents from the two different versions.
You can use the "File > Share" menu item in Skyline to create a .zip file containing the Skyline document and supporting files. Files which are less than 50MB can be attached to this support request. You can upload larger file here:
https://skyline.ms/files.url
-- Nick

We have used a range of test cases to evaluate the different versions - 21.1 vs 22.2. We did not find any differences in the quantitative values we obtained. Of course it is not practical to go through all possible scenarios so we also try to evaluate each feature update to determine if it could impact our result