Hello,

I am currently designing a protein binder targeting a kinase, and I’m using Rosetta for structure-based filtering. I received a PDB template binder from my professor and have been generating variants, docking them to the target, and evaluating the resulting complexes.

Initially, I used the total score of the receptor–binder complex after docking as a filter, but I soon realized that other Rosetta score terms might be more appropriate for evaluating binding affinity. In particular, I came across dG_separated, which I understand estimates the binding free energy between receptor and binder. I’m considering using it as a filter, but I’m not sure what threshold values would be considered meaningful.

While I did some searching, most papers seem to report selecting top candidates based on score ranking without mentioning a specific dG cutoff. This makes it hard to define a clear criterion grounded in literature.

Are there any recommended or empirically supported cutoff values (e.g., < –20, < –30)?Or, is there a more appropriate metric or strategy I should consider?

Any insight from those experienced with Rosetta binder design would be greatly appreciated. I plan to express and experimentally validate selected designs, so I want to be as rigorous as possible at the computational stage.

Thank you!