Is the "model-default" script advisable for mutants with high structural
variations?
Regular comparative modeling, by construction, will yield models that
are structurally similar to the templates, so if you are expecting a
large structural change, you won't see it using "model-default". The
script in the wiki is not using distance-based (homology-derived)
restraints, and allows the SNP to relax using only CHARMM and dihedral
forcefield terms.