My protein of interest has ATP and Mg2+ in its active site. In crystal structure, ADP and BeF3 are present in place of ATP. Is it acceptable to build ATP in PyMol out of ADP and BeF3?
PyMOL does not have any tools to regularize bond lengths and angles, or to minimize the structure. You might take an ATP structure from another source, fit the base on that of the ADP and change the torsion angle to adapt the conformation of the ARP to the binding site - However, it will be easier to build the model in a proper molecular modeling program rather than tryin to force a molecule viewer to do a task that exceeds its capability.
You might want to try just removing ADP and BeF3 from the pdb file and the docking ATP. I find docking with well defined binding pockets and substrates tends to be pretty accurate. Depending on what you are using this model for, you may also want to avoid modeling the structure by hand or you can mess up the energetics of the system.