antibody binding inside structure of the target

[keithgmitchell]

Hi I love the tool! But i am having some trouble with my antibody and the produced design being inside of the target. Do you have any advice for this?

[kxz18]

Hi! Thanks for your interest in this work! I think this is a limitation of the current method as I didn't set any features indicating the surface orientation of the epitope. If different random starting states cannot solve the problem, I suggest changing the definition of the epitope (e.g. adding or removing several residues) to induce changing in the outputs. If you are doing research on the ML algorithms, it might also be interesting to explore possible ways to elliminate this problem, such as clash loss, additional surface embeddings and so on.

[github-actions[bot]]

This issue is stale because it has been open for 30 days with no activity.