The advantage of this algorithm is the independence from the type of structure you want to be canonically numbered.
This means, that you do not have to change the algorithm just because the description of your molecule changed. That applies not
only to changes done in the topological (2D) description of the molecule but also extends to the possibility of canonically
numbering 3D structures. The only thing to change is the kind of description that allows to determine the difference between two
atoms in a spatial representation of the molecule.
No matter, whether you describe the 3D structure only by tetrahedral angles or add additional information like stereochemically
interesting properties, all those descriptors will allow an additional refinement of the initial classification so if the
computation of those properties from a given 3D structure does not take too much time, you may even be able to compute the 3D
canonical numbering faster than the 2D numbering since you start with a better initial classification.
Even if not all properties available can be applied for building the initial classification, they may be used during the iterated
classification for refining the classes of atoms with respect to a class (containing only one atom) from the memorized class list.
The backtracking algorithm is totally independent from those descriptions.
© Chr. Benecke, Oct. 1995