Number of peptides needed to capture 1%, 5%, 10% or 50% of epitopes detected by mass spectrometry.

The lower the number of peptides needed to capture the respective number of epitopes, the better the performance of the prediction model. This is also illustrated by the percentage of false identification (false positive rate, FPR) reported here. Peptides were rank ordered according to regression scores, for a total of 512,423 unique peptides and 8,807 hits. Of note, only the maximal transcript expression was used for peptides with multiple potential origins.