How to use seq_ppi for viral–human PPI binding affinity prediction from sequences?

[BIN-PS]

Hello,

Thank you for making this great tool available! I am currently interested in predicting binding affinity strength between viral and human proteins, given only their amino acid sequences.

I went through the documentation and examples, but I have a few doubts:

Can seq_ppi directly output binding affinity values (e.g., KD, ΔG, or some score) for a given pair of protein sequences, or does it only predict binary interactions (yes/no)?

If affinity prediction is supported, could you please guide me on how to prepare the input dataset?

If not directly supported, would you recommend a modification or extension of the framework to adapt it for affinity regression tasks?

My goal is to build a viral–human PPI binding affinity predictor, so any guidance on the best way to adapt seq_ppi (or pretrained models) for this use case would be greatly appreciated.