Here are a few suggestions:
1. Watch the most recent webinars on Large-Scale DIA:
http://skyline.ms/webinar14.url
http://skyline.ms/webinar15.url
2. Changes to your Transition Settings
- Filter tab - change To = "Last ion - 1" to "Last ion"
- Filter tab - uncheck "N-terminal to proline"
- Library tab - use Pick = "6" product ions
3. Use Edit > Refine > Advanced and at least "Min transitions per precursor" = "4" (if not 5 or 6)
- This cuts the number of peptides in your document from 2,538 to 928 (or 705) giving you some insight into the spectral quality in your library (not great)
4. Ideally, you would set up iRT retention time prediction
5. You want to add decoys for training your model, which it looks like you did at the time, but they are no longer in the document
It looks to me like you have a fair number of good peaks. If you can't train iRT retention time prediction, you might want to consider a wider RT window like +/- 7 minutes and see whether that improves your detection rate.
Also, it is probably a good idea to test on multiple replicates where you expect a constant concentration and make sure you get consistent detection and low CVs across your runs.
Hope these suggestions help. Thanks for your interest in using Skyline.
--Brendan