Clarification needed on thresholds for residue classification in EquiPNAS predictions

[Biocomputationalist]

Hi,

I’m using EquiPNAS to predict residue-level binding propensities, and I noticed that the no thresholds are provided to classify whether a residue is considered “binding” or not.

Could you clarify:

1. Are there recommended threshold values for each type of binding (DNA, RNA)?

If not, is it acceptable to use a generic threshold (e.g., 0.5) or should thresholds be determined differently for each type of binding?

This information would help in interpreting the predictions consistently.

Thanks in advance!