Compare similarity of protein variants

Question

[GP]

I have some .pdb files with structures of variants of the same protein and I'm looking for a way to quantify how similar (or dissimilar) they are from a reference protein.

I've tried superimposing them and calculating RMSD with ChimeraX or PyMol, the problem I'm having is that some variants are shorter than the reference (truncated proteins) and I would like to take this into account, so comparing the common residues is not a good option for me.

What would be the best way to quantify how different the protein structures are?

[Joe]

You could try another metric like TM-score (as used by the ITASSER server). It does a similar thing to RMSD, but in my experience 'breaks' less easily in the edge cases like you describe.

[dthorbur]

You could also look at foldseek, it's akin to blast but for 3d structures and uses pdbs as input for both database generation and search queries.

I doubt it would be anywhere near as accurate as TM-score, as suggested, but if you have lots of comparisons it would be a very useful and fast filter to remove obvious mismatches. Though it only gives you a score on similarity likelihood, and not dissimilarity. You could also filter results to only include matches with +n% of query and target sequence to reduce partial hits if required.