Going through the paper by Elizabeth Kellog et al in Proteins, i realized that the standard reference energies are not well optimized for Delta-deltaG calculation but more suited for sequence recovery. However, it is stated that a new set of opt-E weights yields better correlation. Can any one help me out in figuring where these new set of reference weights exists and how to incorporate them in the standard score12 function.
I can seem to find any mention of what those weights actually are, either in the paper or in the codebase.
To actually apply them, you would make a copy of the appropriate weights file (e.g. rosetta_database/scoring/weights/score12_full.wts), and then modify the “METHOD_WEIGHTS ref” line. The numbers following are the reference energies for each amino acid type (in alphabetical one letter code order, i.e. A C D E …)