Mike White wrote: > I have been modeling a homopentamer which should have a disulfide bond > in each subunit in the same position as the multiple templates that we > are using (the alignments have the cys residues of the target lined up > with the disulfide-forming cys in the templates). When I carry out the > alignment of the templates using salign, the XX_fit.pdb files for the > templates do not have the disulfides indicated using SSBOND, even though > the original pdb files (XX.pdb) do have the SSBOND lines in them.

Modeller only reads the ATOM and HETATM records from PDB files - any other record (including SSBOND) is discarded. (The only exception is if you turn on salign's write_whole_pdb option, in which case all of the header information from the original PDB file is preserved - only the coordinates are transformed - although Modeller does not "understand" any of that information.)

> I know that automodel will automatically include disulfide restraints > if the templates have them. However,the returned models, while > superimposing nicely on the template structure, do not have the > disulfides, even though the cysteines in the model are right where the > disulfide-forming cysteines are in the template. I assume that automodel > uses the XX_fit.pdb files for the fitting routines, and the fact that > the XX_fit.pdb files are missing the SSBOND lines is why the disulfides > are present in the model.

No - automodel does not read (or write) SSBOND records either. It determines template disulfides using a distance criterion. I'm not sure what you mean by "uses the XX_fit.pdb files for the fitting routines" because automodel does not do any kind of template superposition by default (although you can turn on initial_malign3d).

> Alternatively, I guess I could use model.patch as per node24 in the > online manual to force the disulfides to be where I want them to be, but > this just doesn't seem right if the templates all have the disulfide and > the alignments have the cysteines where they should be.

Well, it sounds like automodel has already put in the disulfides for you ("the cysteines in the model are right where the disulfide-forming cysteines are in the template") so using patch() won't add anything. You could manually put SSBOND records into your generated models if your PDB viewer software requires them, but because Modeller does not read them, you'd be wasting your time trying to introduce them into any of Modeller's inputs.

Ben Webb, Modeller Caretaker