I'm working on a protein that appears to have some aromatic groups stacking in the x-ray structure, but I could not see such conformers in my models(Ca RMSD 1.1 A to X-ray structure) built with rosetta.
I found there is this term "fa_plane" in the scoring weights that is supposed to describe "pi-pi interaction between aromatic groups, by default = 0", however, there's no scoring method defined for it in the source code. So if I change the weight of this term, it will only give an error message: "ERROR: Requested ScoreType fa_plane does not have a registered EnergyMethodCreator."
My question is: does anyone know how to score the aromatic group interactions?
fa_plane doesn't exist! Who knew! (Yes, I am as aggrieved as you). It existed in a personal branch of Rosetta++ but never got checked in to main ++ or ported to Rosetta3.
Steven Combs is working on adding orbitals to Rosetta, which would work for your purpose. I have drawn his attention to this post, but I think that his code isn't released in 3.4.
HackAro also exists, which has this comment:
// Trying to get more "T-shaped" edge-to-face interactions of aromatic side chains.
// This is very similar to Kira's fa_plane in rosetta++, though I haven't bothered
// to parameterize as carefully. This is just a hacky functional form whose derivatives
// are easy to calculate and smooth.
// Perhaps the best approach would be to calculate, e.g., benzene-benzene interactions
// with high level quantum calculations -- but note that you need really careful treatment
// of electron correlations to get dispersion right -- still not quite feasible for all rigid body
// (I also tried using hackelec_aro_aro, i.e., coulomb's law between positively charged aromatic
// hydrogens and the negative partial charges on aromatic carbons -- it didn't quite give me
// what I wanted). Rhiju, Dec. 2009.
You (hopefully) can activate it by giving a nonzero weight to hack_aro.
I have implemented a scoring function in Rosetta to capture aromatic interactions; however, like Steven said, it is not in Rosetta3.4. The next release of Rosetta will have the new score function in it. The score function will be described in the Plos 1 special edition of Rosetta.
Are you trying to work on folding?
I'm working on folding a membrane protein and then docking into homo-multimer. The models with the different aromatic conformers have same backbone conformation and only differ by 1.5 rosetta energy unit in energy. The "correct" conformer does pack better in the multimer as the armotic group in the "wrong" conformation will block a docking site and it couldn't be fixed by "relax"ing the multimer. So the only solution seems to be finding out the "correct" conformer in the monomer stage.
If you have a model with the desired rotamer, can you fix that rotamer with NATRO in the resfile? Or, if the backbones are the same, just paste it in with a text editor and repack (using NATRO in a resfile) with fixbb, if the executable you're using ignores resfiles?