It is a tcsh script, so just make it executable and run with no arguments to get the command-line options. Basically, you just give it two PDB files, and perhaps a one-word space group. The first PDB file is taken to be the "right answer". By default it breaks both files up into "chains" and compares each pair of chains in turn. You can make each file behave as one "chain" by putting the word "nochains" on the command line. By default, I use RMSD as a "score", but if you put the word "correlate" on the command line, it will use electron density map correlation instead. Symmetry operations are resolved by doing the correlations in P1. and I resolve origin shifts with a map deconvolution algorithm, but cphasematch might work better.
The output file is called "neworigin.pdb". By default, the chain IDs in the "wrong origin" PDB file are re-assigned to those of the best-matching "right origin" pdb file (unless "nochains" was specified). The script will also generate a "newlabel_neworigin.pdb" where I try to re-name all the atoms in the "wrong" PDB to match the "right" pdb, and the occupancy column then encodes the by-residue map correlation. This file is more useful for matching up heavy atom constellations.
It ain't elegant. It ain't fast. But I'm pretty sure it works.