[phenixbb] Excluding atoms from coordinate refinement ignored

[Engin Özkan]

Hi Phenix developers,

I have an issue using 1.8.1 (1168) with excluding atoms from coordinate 
refinement. I have defined a certain residue to be excluded from 
coordinate refinement, as can be seen in the .eff file produced by this 
run.

   refine {
     strategy = *individual_sites individual_sites_real_space rigid_body \
                *individual_adp group_adp *tls *occupancies group_anomalous
     sites {
       individual = not (resseq 60 and resname PHE)
       torsion_angles = not (resseq 60 and resname PHE)
       rigid_body = None
...

However, this gets ignored, and this residue and its NCS-related 
brethren get moved into nondescript electron density on a symmetry axis. 
This residue was just fine in an earlier version of Phenix, but when I 
switched to 1.8.1, it started moving into this density. So, I thought 
excluding it from xyz refinement, which is a dirty trick in my opinion, 
would fix this problem. Not so.

I have the feeling that this is due to rotamer corrections as part of 
the torsion NCS implemented in 1.8.1 now. Sticking a zero occupancy 
dummy atom in the density did prevent this residue from moving in. 
Turning off torsion NCS also fixed it. By the way, I have "fix rotamer 
sidechains" off, as I am yet to see it improve things (at medium or low 
resolution).

Thanks and happy holidays!
Engin