Joaquim Rui Rodrigues wrote: > I made some changes to the script examples/automodel/loop.py to do > loop refinement with dope_loopmodel in the > presence of ligands. While this used to work well with modeller 9v4, > with 9v5 I am getting the following error:

Actually, it didn't work with Modeller 9v4. ;) 9v4 would happily run with zero GB/SA radii, but this would result in infinite terms creeping into the scoring function, which would give unpredictable results. So in 9v5 an explicit check is made for sane radii.

> The atom is an MG ion. From solv.lib: > # RSOL GBSA ASP ASPW SCKS OONS JRF > MG 0.000 0.117 0.000 0.000 0.000 0.000 0.000 ! Magnesium Ion > > I suppose I should change the value of RSOL. Is there any recomended > value?

That is exactly right - RSOL is the value that needs to be changed. But if we had a recommended value, it would be in there already. ;) I suggest you check in the literature and perhaps in other MM packages such as CHARMM.

> I foresee more problems for other atom types as solv.lib seems to > have radii data only for protein atom types...

Yes, this is true, because that's all we've used GB/SA for ourselves.

Ben Webb, Modeller Caretaker