Section 3.3.2.1.1. Connectivity tables

The simplest means of storing connectivity information is by means of tables in which, for each atom, a list of indices of other atoms to which it is connected is stored. This approach is quite general; it may serve any type of molecular structure and permits structures to be traversed in a variety of ways. In this form, however, it is extravagant on storage because every connection is stored twice, once at each of the nodes it connects. It may, however, provide the starting material for the algorithm of Section 3.3.1.5.2 and its generality may justify its expense.

From such a list, lists of bonds, bond angles and dihedral angles may readily be derived in which each entry points to two, three or four atoms in the atom list. Lists of these three types form the basis of procedures which adjust the shape of a molecule to reduce its estimated potential energy (Levitt & Lifson, 1969; Levitt, 1974), and of search-and-retrieval techniques (Allen et al., 1979).

Katz & Levinthal (1972) discuss the explicit specification of structural connectivity in terms of a tree structure in which, for each atom, is stored a single pointer to the connected atom nearer to the root, virtual atoms being used to allow ring structures to be treated as trees. An algorithm is also presented which allows such a tree specification to be redetermined if an atom in the tree is newly chosen as the root atom or if the tree itself is modified.

Cohen et al. (1981) have developed methods of handling connectivity in complicated fused- and bridged-ring systems.