Questions about the epitope prediction threshold

[aggelos-michael-papadopoulos]

As we can see in the bibliography the common threshold used in the epitope prediction experiments is 4.5 Armstrongs (see PECAN, EpiPred). On your paper you seem to use 2 Armstrong as a threshold but on the comparison results at table 3 you compare your results with those of 4.5 Armstong. So what threshold value did you use as threshold for the epitope experiment?