docking against a specific pocket

[adrienchaton]

Hi @gcorso and thanks for sharing this great model.

It doesn't seem that DiffDock can take a constrain on the target pocket to dock, as other methods which expect e.g. coordinates for a box in which we expect to dock the ligand. Would it make sense to extract the receptor pocket into a separate .pdb and use this as input for docking?

I see two possible pitfalls

• lacking the global protein context, to me this doesnt seem too bad though
• creating false interfaces to dock against, since the rest of the protein gets cut, some residues may be exposed in the pocket pdb whereas in the full protein they are not

Do you have any thoughts on that or tips on how to constrain DiffDock to a known pocket of my receptor please?