I am using kic loop protocol to refine a big loop with 12 aa with rosetta. I noticed that the energy is positive from generated .PDB file:
Shouldn't it be a negative value in principle?
BTW, how can we tell Rosetta treat SS bridge in the latest version? I would suggest why not add an option to Rosetta so that it can treat SS bond automatically according to SS atoms distance. Many software such as: Modeller, Gromacs, Amber can do this automatically. It shouldn't be so difficult to achieve this function.
thx a lot
A positive energy means the model is bad in some sense. Usually it means there are clashes, as the repulsive term increases in energy fastest and accrues the largest positive values. It might nor might not be problems in the loop you modeled: a clash on the other side of the protein still results in a positive score.
The general solution is to determine if the scores of the residues in/near the loop are good, and if so, accept that the model of the loop is plausible. If you want a globally better score, relax the protein before or after loop modeling to fix small global errors.
Rosetta should detect disulfides automatically. Where is it not doing so?
I don't think the overall structure is bad. It is a crystal structure and I minimized the initial structure before loop KIC model. I generated over 50 K loop models in all, and to my great surprise that all of them has a positive total energy values.
How strange it is, I never see any model from my KIC generated models treat SS bond correctly. I protonated all my Cys automatically although I visualize my initial structure in VMD, PYMOL, Chimera and maestro and they do form SS bond. Shall we rename the Cys to a special name so that Rosetta can handle it correctly?
I'd recommend taking a look at the per-residue energies in the structures. (Either in the table at the bottom of the output PDB, or by using one of the analysis applications (https://www.rosettacommons.org/docs/latest/residue-energy-breakdown.html). On some structures there are residues which are bad for some reason or another, and so might hurt your overall energy, even though the rest of the energies are fine.
Disulfides are an issue. Rosetta has some internal disulfide detection logic, and it's sometimes hard to override it. There is a commandline flag "-fix_disulf" which take a filename of a file which has a list of pose-numbered disulfide bonded residues, one disulfide bond per line that you can try, although not all protocols obey it. Also pay attention to if you have any other disulfide-flags set.
Note that if you have a disulfide that's being modeled as two protonated cysteines, the hydrogens on the cysteines will likely clash, giving you a bad, positive score. (Having protons present in the input structure may also inhibit Rosetta's automatic disulfide detection scheme.)