Open Source Molecular Symmetry Perception Tools

Question

[Nick Vandewiele]

Hi,

I am looking for a tool that can perceive the molecular symmetry of a given structure without having to specify/generate the 3D coordinates. With molecular symmetry i mean the external symmetry number of chemical species.

Do you know a tool that does this?

regards,

Nick

[Gilleain]

I'm not sure what you mean by "external symmetry number" - could you give a short example?

[Noel O'Boyle]

Do you mean graph (aka topological) symmetry? The term molecular symmetry usually refers to the arrangement of the 3D atoms in space.

[Gilleain]

Ah. After some googling, it appears to be something like the order (size) of the subgroup of rigid (automorphic?) permutations of the atoms. The only paper I can find is this : linkinghub.elsevier.com/retrieve/pii/009784859180020M

[Nick Vandewiele]

i mean topological symmetry. E.g. benzene has symmetry number 12. Gilleain's description is kinda spot on. It is a contribution in the entropy of a molecular species. There are indeed not a lot of papers about it...

Answer 1

[Egon Willighagen]

The Chemistry Development Kit has algorithm implementations to do this. I am no expert on this, but symmetry detection is an old but still hot topic and I recommend browsing Gilleain's blog on involved matters (search for SMSD and signatures). Developers of OpenBabel are also looking into relevant algorithms, and Tim's blog contains a few interesting posts.

Answer 2

[Noel O'Boyle]

You can find the number of automorphic permutations using the Open Babel library. See FindAutomorphisms in the Open Babel API. You'll have to write a small C++ program or else you can access it from Python: [?][?]import pybel mol = pybel.readstring("smi", "c1ccccc1Br") mappings = pybel.ob.vvpairUIntUInt() success = pybel.ob.FindAutomorphisms(mol.OBMol, mappings) print "Number of mappings is %d" % len(mappings) for mapping in mappings: print mapping [?][?] Output is: [?][?]Number of mappings is 2 ((0, 0), (1, 1), (2, 2), (3, 3), (4, 4), (5, 5), (6, 6)) ((0, 4), (1, 3), (2, 2), (3, 1), (4, 0), (5, 5), (6, 6))[?][?]

[Nick Vandewiele]

Noel,

thanks for the tip. the automorphism group of a graph is the concept that I was looking for, I think.

Answer 3

[Nick Vandewiele]

I found Gilleain's document on Graphs, Groups, Symmetry in CDK, which explains a lot of the things I am looking for.

So, I know that CDK is able to generate symmetry classes for atoms in a molecule (using the MoleculeSignature.calculateOrbits() method). So, this gives me the partitioned sets of atoms that are topologically equivalent.

However, this is not the automorphism group of the molecule, which I am looking for. From Gilleain's document, automorphism group generation is already done through the InChI-c library but not available in CDK as a method as such.

Am I right in saying that CDK currently does not have an automorphism discovery algorithm (e.g. ported from nauty) implemented as such?

btw: maybe this question ought to be posted in cdk-user mailing list or so...

[Gilleain]

I'm a little embarrassed about the unfinished nature of that document, but I hope it's useful :)

So, no there is not a way that I know of to get the automorphism group from the inchi. It can give you the canonical labelling (from the AuxInfo) but that's not what's required here.

There is code here : https://github.com/gilleain/cdk_signature/tree/master/src/org/openscience/cdk/group that is a not-yet integrated cdk module. It has an SSPermutationGroup class and a DiscretePartitionRefiner. If the refiner is given an initial partition based on signatures, it should give the correct aut group.