Hi Stoyan,
We use the most appropriate probability value supplied by the search pipeline. When a q value or FDR estimate is available, we try to use that, but some search pipelines provide only posterior error probability or an expectation score. I was recently at ETH teaching the DIA/SWATH course, in which the tutorials used Comet with the TPP. The TPP is one of the pipelines which provides a posterior error probability, and participants were taught to use a tool called Mayu to estimate the PEP with a 1% protein-level FDR. Once that was estimated, then they entered it as the cut-off for Skyline. If your Protein Pilot pipeline uses a non-FDR related probability score, but you have a tool for estimating FDR, then you would just enter the appropriate cut-off for the FDR you want. (Note: if the score is one where zero is best, then you use 1-score as the cutoff. Skyline will flip the value appropriately.)
If you have a pipeline, like MaxQuant or Spectrum Mill where you can apply a cut-off before giving the results to Skyline, then you can just enter 0 in the score cut-off, which explicitly tells Skyline to ignore the cutoff. Since in all other cases, this is a very bad idea, we make you explicitly enter zero to indicate you know what you are doing and you have already pre-filtered your spectrum matches.
Hope this helps. Thanks for using Skyline in your research.
--Brendan