How similar are these two structures, and where do they differ?
Compare two protein structures
Superposition, TM-score against RMSD, and reading the difference honestly.
Decide first what question you are asking, because the two common metrics answer different ones. RMSD is an average deviation over the residues you paired, in ångströms. It is the right measure for two copies of the same protein — a predicted model against its experimental structure, or the same protein with and without a ligand.
RMSD is a poor measure of whether two different proteins share a fold, for two reasons. It grows with length, so 4 Å across 60 residues and across 600 mean quite different things; and one flexible loop swinging away drags the average as hard as a genuinely wrong topology. TM-score was built to fix both: it normalises by length and saturates, so distant residues stop adding penalty. Above about 0.5 two structures share a fold; below about 0.3 they are no more alike than two proteins picked at random.
A single number of either kind hides where the difference is, and that is usually the interesting part. A per-residue deviation plot separates the case of two structures differing everywhere a little from the far more common one of a rigid core with one hinge or one loop moving — which is a result, not noise.
TM-score is asymmetric: it depends which structure’s length you divide by. A small domain matching part of a large protein scores high one way and low the other, and both numbers are correct. Quote the normalisation along with the score.