[Molecularmechanics] Recycling CML

[Konrad Hinsen]

On Tuesday 11 November 2003 09:27, Peter Murray-Rust wrote:

> Fine - that is the cleanest way to go. Things I have to remember for CML
> are: - the atoms involved in the new bond  will generate the need for
> forcefield parameters for them and their immediate bonds

Yes, indeed.

> - can the new bond be multiple? Could there be two bonds to the free atom?
> These cases are rare and can probably be ignored first time round.

I can't cite any example but I don't see any problem with such a situation if 
it should occur.

> - does the new bond formation generate new stereochemistry? This could
> happen with (say) 3-coordinate S as in s_adenosyl methionine. Again rare,
> but common enough in regular chemistry

Perhaps. I don't worry about such problems simply because I see fragments as 
nothing more than abbreviations in the specification of a molecule. I would 
always do assignments of bond order, force field parameters, stereochemical 
properties, etc., on a whole molecule.

> >I think that's the only reasonable approach, because otherwise any program
> Do you rely on the directionality of the ligand atoms to determine the
> stereochemical parameters of the atoms in the bond. e.g. if you take formyl
> (HC(=O)-) forming HC(=O)-X does the geometry of the C atom remain
> independent of the nature of X? If X == NHR then there is a 2-fold torsion,
> if CH3 a 6-fold one.

Again, I would do all assignments and interpretations in the end, not on the 
fragments.

> I am encouraged by your model and have been thinking about its
> implications. Essentially we have
>
> <molecule id="m1">
>    <molecule ref="acetyl"/>
>    <molecule ref="nmethyl"/>
>    <bondArray atomRef1="acetyl:c1" atomRef2="nmethyl:n1"/ order="1"/>
> </molecule>
>
> This is legal but the semantics of this are not absolutely explicit in CML
> at present. It should result in a new molecule (m1) with a single
> connection table.

Yes, that is the intention.

> It would be allowable to create atomSets which described
> the original fragments, e.g.
>
> m1.getAtomSet("acetyl") could return all the acetyl atoms and
> m1.getBondSet("acetyl") the bonds. It is then up to the implementer to

Yes, of course, this is how my code works - but it shouldn't matter for the 
definition of a file format.

> retain the information. There could also be an atomSet describing the new
> peptide bond - note that atomSets can have overlapping membership -
> molecules cannot.

I have never seen a use for such an atomSet, but it would be trivial to 
construct.

> If this makes sense I am happy to write it up more formally.

It makes sense to me at least!

Konrad.
-- 
-------------------------------------------------------------------------------
Konrad Hinsen                            | E-Mail: hinsen@cnrs-orleans.fr
Centre de Biophysique Moleculaire (CNRS) | Tel.: +33-2.38.25.56.24
Rue Charles Sadron                       | Fax:  +33-2.38.63.15.17
45071 Orleans Cedex 2                    | Deutsch/Esperanto/English/
France                                   | Nederlands/Francais
-------------------------------------------------------------------------------