In BiblioSpec, we switched from using the identity threshold to the homology threshold based on this information from John Cottrell at Matrix Science:
"Mascot has two score thresholds - the identity threshold and the homology threshold. We recommend using the homology threshold in almost all cases because it will give best sensitivity for a given FDR, and the Protein family summary reports expect values based on the homology threshold. I suspect Skyline is using expect values based on the identity threshold. You can confirm this by switching to the older, Select summary report and seeing whether the lists of peptides then match up"
Following this change, a user requested the use of the identity threshold: https://skyline.gs.washington.edu/labkey/announcements/home/support/thread.view?rowId=18891
We should determine if/how to support use of the old identity threshold.