[phenixbb] residue selection for rsrl?

I am refining a structure at 2.9 Å resolution and have two side chains that keep moving out of density during reciprocal space refinement; both side chains should fit the resulting Fo-Fc density in favored rotamers. Turning on real-space refinement fixes this problem but leads to higher Rfree and poorer geometry overall. Can I select a subset of residues for real space refinement while still performing reciprocal space refinement on the whole asymmetric unit? Thanks for any help Kevin