data structure design question

[Konrad Hinsen]

"Andrew Dalke" <dalke at acm.org> writes:

> mention bond types.  I don't have MMTK handy, but
> it is definitely atom type-based, and I don't think
> it has a definition for bond type.  The most common

MMTK defines atom and bond objects, and lets you add whatever chemical
attributes you want to them via the MMTK database. For the standard
force fields (Amber etc.), this means atom types but nothing for
bonds. I would be no problem to have bond attributes as well, but I
haven't yet seen a need for them. But maybe that's just because I am
not a chemist either.

I suspect that the atom type approach is more general, since atoms are
a much better defined concept than bonds. Ultimately, bonds are just
particularly strong interactions between two atoms. Interactions exist
in all kinds of strengths, so any definition of a bond or bond order
leaves some cases open for interpretation.

Conclusion: there is no one correct computer model for molecules.
It all depends on what you want to do with it.
-- 
-------------------------------------------------------------------------------
Konrad Hinsen                            | E-Mail: hinsen at cnrs-orleans.fr
Centre de Biophysique Moleculaire (CNRS) | Tel.: +33-2.38.25.55.69
Rue Charles Sadron                       | Fax:  +33-2.38.63.15.17
45071 Orleans Cedex 2                    | Deutsch/Esperanto/English/
France                                   | Nederlands/Francais
-------------------------------------------------------------------------------