Dear all,

I have been stuck in the problem for a while. I appreciate any suggestions.

I have two zinc binding domain in my 3A structure which are Cys4 type.  The geometry is OK for one domain, but it is not refined properly in the other one.

The problems during phenix refinement are
1.  Zinc is always refined to position ~0.5A off from its correct position
2. The refined bond lengths of four Cys-Zn are all below 2.3
3. Sometimes, distances between SG atoms of nearby cysteines are too close, and coot just connect them as disulfate bond.
4. Sometimes, I get the following message showing that all bond angles are outliers with >8 sigmas, I listed part of the outliers.
        
Top angle outliers (sorted by deviation):
atoms                                 ideal       model  difference   deviation (sigma)
----------------------------------------------------------------------------
 SG  CYS D1204            57.894     104.470      46.576                9.32
ZN    ZN D2001      
 SG  CYS D1112      

  SG  CYS D1201            50.842      92.496      41.654                8.33
ZN    ZN D2001      
 SG  CYS D1204      

 SG  CYS D1204           126.952      92.496     -34.455               -6.89
ZN    ZN D2001      
 SG  CYS D1201      

 SG  CYS D1201            52.183      86.105      33.922                6.78
ZN    ZN D2001      
 SG  CYS D1112      

 SG  CYS D1194           152.275     119.729     -32.546               -6.51
ZN    ZN D2001      
 SG  CYS D1112      

 SG  CYS D1112            89.273     119.729      30.456                6.09
ZN    ZN D2001      
 SG  CYS D1194      

I also post my metal.edit file for refinement 

refinement.geometry_restraints.edits {
  bond {
    action = *add
    atom_selection_1 = name ZN   and chain D and resname ZN and resseq 2001
    atom_selection_2 = name  SG  and chain D and resname CYS and resseq 1201
    distance_ideal = 2.310000
    sigma = 0.100
  }
}

refinement.geometry_restraints.edits {
  bond {
    action = *add
    atom_selection_1 = name ZN   and chain D and resname ZN and resseq 2001
    atom_selection_2 = name  SG  and chain D and resname CYS and resseq 1194
    distance_ideal = 2.310000
    sigma = 0.100
  }
}

refinement.geometry_restraints.edits {
  bond {
    action = *add
    atom_selection_1 = name ZN   and chain D and resname ZN and resseq 2001
    atom_selection_2 = name  SG  and chain D and resname CYS and resseq 1112
    distance_ideal = 2.310000
    sigma = 0.100
  }
}

refinement.geometry_restraints.edits {
  bond {
    action = *add
    atom_selection_1 = name ZN   and chain D and resname ZN and resseq 2001
    atom_selection_2 = name  SG  and chain D and resname CYS and resseq 1204
    distance_ideal = 2.310000
    sigma = 0.100
  }
}

refinement.geometry_restraints.edits {
  angle {
    action = *add
    atom_selection_1 = name  SG  and chain D and resname CYS and resseq 1194
    atom_selection_2 = name ZN   and chain D and resname ZN and resseq 2001
    atom_selection_3 = name  SG  and chain D and resname CYS and resseq 1201
    angle_ideal = 112.172178
    sigma = 5
  }
}

refinement.geometry_restraints.edits {
  angle {
    action = *add
    atom_selection_1 = name  SG  and chain D and resname CYS and resseq 1112
    atom_selection_2 = name ZN   and chain D and resname ZN and resseq 2001
    atom_selection_3 = name  SG  and chain D and resname CYS and resseq 1201
    angle_ideal = 87.092184
    sigma = 5
  }
}

refinement.geometry_restraints.edits {
  angle {
    action = *add
    atom_selection_1 = name  SG  and chain D and resname CYS and resseq 1112
    atom_selection_2 = name ZN   and chain D and resname ZN and resseq 2001
    atom_selection_3 = name  SG  and chain D and resname CYS and resseq 1194
    angle_ideal = 118.116518
    sigma = 5
  }
}

refinement.geometry_restraints.edits {
  angle {
    action = *add
    atom_selection_1 = name  SG  and chain D and resname CYS and resseq 1204
    atom_selection_2 = name ZN   and chain D and resname ZN and resseq 2001
    atom_selection_3 = name  SG  and chain D and resname CYS and resseq 1201
    angle_ideal = 91.677138
    sigma = 5
  }
}

refinement.geometry_restraints.edits {
  angle {
    action = *add
    atom_selection_1 = name  SG  and chain D and resname CYS and resseq 1204
    atom_selection_2 = name ZN   and chain D and resname ZN and resseq 2001
    atom_selection_3 = name  SG  and chain D and resname CYS and resseq 1194
    angle_ideal = 130.829256
    sigma = 5
  }
}

refinement.geometry_restraints.edits {
  angle {
    action = *add
    atom_selection_1 = name  SG  and chain D and resname CYS and resseq 1204
    atom_selection_2 = name ZN   and chain D and resname ZN and resseq 2001
    atom_selection_3 = name  SG  and chain D and resname CYS and resseq 1112
    angle_ideal = 104.886672
    sigma = 5
  }
}


Questions, 
1. Why the ideal bond angle are different in the metal.edit file and final model ? How does phenix determine the ideal bond_angle?
    For example, in the metal.edit file
  atom_selection_1 = name  SG  and chain D and resname CYS and resseq 1204
    atom_selection_2 = name ZN   and chain D and resname ZN and resseq 2001
    atom_selection_3 = name  SG  and chain D and resname CYS and resseq 1112
 The angle_ideal = 104.886672 

But in the final model
 
atoms                                 ideal       model  difference   deviation (sigma)
----------------------------------------------------------------------------
 SG  CYS D1204            57.894     104.470      46.576                9.32
ZN    ZN D2001      
 SG  CYS D1112      

2. Do I need to put the following LINK in the pdb header to run refinement?
LINK        ZN    ZN D2001                 SG  CYS D1201     1555   1555  2.34  
LINK        ZN    ZN D2001                 SG  CYS D1112     1555   1555  2.33  
LINK        ZN    ZN D2001                 SG  CYS D1194     1555   1555  2.38  
LINK        ZN    ZN D2001                 SG  CYS D1204     1555   1555  2.37

3. After phenix refinement, the output pdb doesn't contain these LINK lines no matter what the inpout pdb is. Does phenix_refine use these restrains of LINK lines?

Thanks.

Yu Zhang
Postdoc fellow
Rutgers University