Joel Tyndall wrote: > I have been building some models with disulfide patches. One of the > disulfides presents a model with basically bad geometry (cys66-cys40) > which is troubling. I have attached model and top file. (I'm using 8.2 > with old style formats). > > Any suggestions as to how to rectify this (fix the geometry)? I'm trying > to gauge which is the appropriate disulfide connectivity between this > model and another and this ain't helping the cause.
You don't say whether all of your models are affected - I see from your script that you're building 5 of them. If only some have the bad geometry, you should just discard those models (filter on the objective function and/or the number of violations). It's not surprising that occasionally you get models with bad geometries, since some of the random starting positions will be frustrated in some way.
If all your models have a bad geometry, building more models may help (5 is not very many). Otherwise a close look at the log file and/or per-residue violations (.V file) may point you to the cause.
Ben Webb, Modeller Caretaker