An important feature of the new algorithm is an automatic adaption of the complexity of the algorithm to the complexity of the
individual molecule. Thus, in most cases the algorithm will get a canonical numbering very fast and only certain cases
will require more effort.
The canonical numbering procedure can be applied to molecules from various sources. The authors use the structure generator MOLGEN as an example source of different molecules, converted to
3D by MM2 energy calculations. MOLGEN requires as input only a
brutto formula, and it can take several restrictions into account. All mathematically feasible molecules with these properties are
then generated in a fast and redundancy-free way and shown on the screen. Additionally, a database module will allow to store and
retrieve selected molecules. The results can be exported to other programs or inserted directly into documentations.
© Chr. Benecke,
Oct. 1995
Abstract
Canonical numberings are indispensable for any chemical database, since they are needed to define the identity between
single molecules. Of course, the identification may be computed with respect to different properties of the molecules, depending
on the representation of the molecule in question. Accordingly one expects also many different numbering rules for all these
cases.
In this article we present a unified approach for a large variety of representations. The algorithm can also be used to study 3-D
placements where bond angles may be of different significance at different positions of the molecule.