Reactions and Transformations - Design and Use of Relational Databases in Chemistry - page 101

1

1

Rxn

C

C

O

[O,Cl] N

O

2 5 3

2

+ +

Rxn

5 A H 4

3 NH 4

Figure 9.1 Atom-mapped reaction of an amine with an acid or acid chloride.

be [C:1]([O,Cl:5])=[O:2].[N:3][H:4]>>[N:3][C:1]=[O:2].[*:5][H:4]. Figure 9.1
shows a depiction of this SMIRKS reaction. Notice that the second prod-
uct, which was water in the initial example, is now expressed as [*:5][H:4].
This [*:5] is necessary because the second product could be either water or
HCl. The use of [H:4] completes the full accounting of every atom involved
in the transformation. Hydrogen atoms not involved in the transforma-
tion need not be explicitly specified. This atom-mapped SMIRKS has also
left out 2 extra carbon atoms, one attached to each reactant in the original
example. These carbons properly belong to the specific reaction SMILES
discussed above, but they do not participate in the transformation and
need not be specified in the SMIRKS.

9.3.1 Unimolecular Transformations
Before considering how SMIRKS can be used to carry out transforma-
tions with multiple reactants, first consider simpler unimolecular trans-
formations. These are discussed separately because of the important use
of unimolecular transformations to enforce the consistent use of SMILES
throughout the database. This improves the integrity of the data in a
chemical sense, rather than a relational database sense as discussed pre-
viously. The root of the issue is this: There are multiple ways to repre-
sent the same molecular structure due to the limitations of valence bond
theory. In valence bond theory, upon which SMILES is based, atoms have
formal charges, most often zero. The bonds between atoms are shared
pairs of electrons and may consist of multiple shared pairs giving rise to
double, triple, or possibly even higher-order bonds between atoms. This
simple theory, while quite powerful and applicable to a majority of chemi-
cal structures, leads to certain ambiguities.
It is generally conceded that simple valence bond theory cannot
adequately explain the bonds between the carbon atoms in benzene.
This classic conundrum is often resolved by stating that there is a sort of