Is it possible to use a known structure (in pdb format) for feature generation and interactions prediction

[ZKrekhno]

I am trying to model interactions with proteins for which the structures are already known (either crystal structures are solved or they have Alphafold predicted structures on Uniprot). Is there a way to just use these pdb files for interactions prediction rather than generating features from basic sequences?

[FreshAirTonight]

You may use PDB structures as templates as part of input features. But the prediction requires sequence input features as well.

[ZKrekhno]

Oh that sounds good. Thank you for the quick reply!