To obtain the restraints on pyruvate, we first superpose the structures of several LDH and MDH enzymes solved with ligands. Such a comparison allows us to identify absolutely conserved electrostatic interactions involving catalytic residues Arg161 and His186 on one hand, and the oxo groups of the lactate and malate ligands on the other hand. The modeling script can now be expanded by creating a new class 'MyModel', which is derived from automodel but which differs in one important respect: the special_restraints routine is redefined to add, to the default restraints, some user defined distance restraints between the conserved atoms of the active site residues and their substrate. In this case, a harmonic upper bound restraint of 3.5±0.1Å is imposed on the distances between the three specified pairs of atoms. A trick is used to prevent MODELLER from automatically calculating distance restraints on the pyruvate-TvLDH complex; the ligand in the 1emd_bs template is moved beyond the upper bound on the ligand-protein distance restraints (i.e., 10).

I am unable to understand the process of moving the ligand in 1emd_bs beyond the upper bound? As in, how does one not let modeller choose the default restraints from the template such that the user-defined restraints are considered for modeling ligand into the target protein.