[modeller_usage] how to fix some residues not lying in allowed regions of Ramachandran plot

Dear All,

I built 200 tetrameric models of my protein using 2 templates. I selected best
model as the one having the least DOPE score. On ssessing the stereochemical
quality of the model, I found that it had 28 outliers residues (phi, psi), that
were lying in disallowed regions of Ramachandran plot. Most of these residues
are functionally important and hence can not be ignored. Will it be fine to
select each of these residues and model them using loop modelling?

Or can anyone please suggest me some other (better) way of fixing such bad
residues?

With Regards

Ruchi