I have been encountered a problem when the ligand in the active site is fitted (it is a modified form of glutamate). After refinement with refmac the one of the oxygen atom bonded with carbon is always getting lost and no coordinate is being available afterwards for this bond only, although cif file of the ligand is used during the refinement. I have used sketcher (ccp4) to get the cif file of the ligand. please suggest what can be done. I tried phenix also to generate the cif from elbow but did not work. 

Thanks in advance

Jeorge