I was using Rosetta 2016.02 "local refinement" to calculate the interface scores of known protein complexes (in a bound form).
Now, I tried docking the same protein complexes, with exactly the same flags with Rosetta 2018.21. I realized that the content of the score file has changed and it gives totally different I_sc and total_score. Almost any protein complex that I tried gives favorable results now (i.e. I_sc < -5). Why is this the case?
I couldn't find what has changed since 2016.02 version.
As an example, I tried local refinement for B and J chains of the attached pdb file (after prepacking). Rosetta 2016.02 (local refinment) gave me total_score of -150 and I_sc of -4.73. However, Rosetta 2018.21 gave me total_score -96 and I_sc -19.5.
I would be very glad if you could help me.
The older RosettaDock used the docking score function originally developed in the mid-2000s. We recently migrated the full atom score funciton to REF2015, which is the default all-atom score function in Rosetta now. We rcommend that you run the docking protocol with the new score function. However, if you need to run with the older score function, please the add the following flags:
-score:weights docking -docking:high_patch docking_min
Hope this helps.
thank you so much Shourya for the explanation!