Hi,


sounds like a bug. Could you please send me PDB file (please send to me directly, not entire mailing list)?


Meanwhile, a quick work-around is to use link_ligands=False (if you are using command line) or if using the GUI:


Refinement settings -> Automatic linking options -> uncheck "Link ligands to protein" .


Pavel



On 7/28/16 05:39, Amber Lee wrote:

Dear all, Dear Pavel,


I have encountered a problem where a new bond was generated between the O of SO4 and O(H) of my ligand (with a glycerol head group, I call it glycerol here).


As shown in the figure in the Dropbox link below, I have placed the glycerol (purple) in the density and it fits well. The distance between the O of SO4 and the OH of glycerol was 2.55 angstrom. After refinement (yellow) however, the glycerol was pulled out of density with the distance changed to 1.27 angstrom, therefore forming a new bond.


When hydrogens were added to the glycerol, the glycerol was again pulled, forming a C-O bond between SO4 and glycerol. 


When the positions (xyz) of the glycerol atoms were fixed, the SO4 was pulled towards glycerol, making a O-O bond between SO4 and glycerol.


When the positions of the SO4 and glycerol were both fixed during the refinement, the result came out saying the bond length was too long between the the O (SO4) and the O (glycerol).


Thus I thought Phenix must treated it as a covalent bond.


I have tried Rafmac and it did not do this to the model. The cif files etc were the same for Phenix and Rafmac.


The problem appeared both on version 1.10-2152 and the latest version 1.10.2155.


How could I fix this problem?


Many thanks,


Jingyi Hu