A substructural analysis method for structure-activity correlation of heterocyclic compounds using Wiswesser line notation.

A method of substructural analysis for structure-property correlation of data sets including heterocyclic structures is described. Structural features allowing representation of occurrence and position of heteroatoms, ring fusions, and substituents are derived automatically from Wiswesser Line Notation representations. Structural feature sets of varying degrees of complexity may be derived, suitable either for sets of derivatives of a single ring system or for mixed sets containing several ring systems. The method is evaluated by correlating pKa values of 169 nitrogen heterocycles, including multisubstituted derivatives of 11 different ring systems. The technique could be carried out automatically with large machine-readable structure-property files and applied to a wide variety of properties.