From: Ashar Malik (
Date: Wed Jan 24 2018 - 01:42:55 CST

Protein structure comparison is a two step process.

Structural alignment followed by Superposing one structure on another based
on the calculated alignment.
Classically the VMD visualiser only superposes equivalent structures which
means that the atom count between compared structures needs to be equal.
If your two structures are A (e.g. 212 residues) and B (e.g. 424 residues)
you e.g. if you know that all of Structure A (or some part thereof) aligns
exactly within Structure B
you can manually align the structures. For instance if from prior
information you know structure A (residue 30 till 40) and structure B
(residue 80 till 90) should be equal then you can load both into VMD

and do something like:

set structure1 [atomselect <structure1ID> "resid 30 to 40 and name CA"]
set structure2 [atomselect <structure2ID> "resid 80 to 90 and name CA"]

The selection of CA's ensures equal numbers of atoms.


measure fit structure1 structure2 --- will then give you a transformation
matrix which you can apply on all of structure A to superpose it onto
structure 2 using the move command in VMD.

Alternatively if you don't know the mapping between the two structures you
have to first calculate that. A popular method is sequence-based where
sequence alignment is used to work out which parts of the structure match
If you want to use only structural information you can use a structural
comparison metric like DALI, CE, TM-Align, Mammoth etc.

These are not packaged with VMD (I think).
When you compare two structures the result from any of these metrics will
also include a transformation metric. You can then import that into VMD
(some TCL code required here) and apply that transformation matric onto the
original query structure and get your superposed states.

Given that your structures differ 212 residues - be careful in the aligning

Hope this was helpful.

On Wed, Jan 24, 2018 at 3:49 PM, Wong Li Zhe <>

> Dear Sir/Madam,
> I have 2 different protein structures and would like to superimpose them.
> They both differ by 212 residues.
> Any insights on how should I superimpose them?
> Best regards,
> Li Zhe