Open xiaoxiao349 opened 5 months ago
Can I use AF2_multimer for validation? I found that use_multimer=True would help with the in silico score.
It would probably be a good idea to update this filter to the AlphaFold-Multimer LIS Score instead of the pAE filtering. It looks like this is significantly better based on their paper's results Enhanced Protein-Protein Interaction Discovery via AlphaFold-Multimer. They also have some examples in the GitHub on how to do it with AlphaFold3 as well.
My experience is to set a big number of the designed binders (around 10K). I could get ~10 satisfied designed binders.
May I ask how to improve pae_interaction, as the pae_interaction I got is always above 10 ?