Numerous quantitative structure−activity relationships (QSARs) have been developed using topostructural, topochemical, and geometrical molecular descriptors. However, few systematic studies have been carried out on the relative effectiveness of these three classes of parameters in predicting properties. We have carried out a systematic analysis of the relative utility of the three types of structural descriptors in developing QSAR models for predicting vapor pressure at STP for a set of 476 diverse chemicals. The hierarchical technique has proven to be useful in illuminating the relationships of different types of molecular description information to physicochemical property and is a useful tool for limiting the number of independent variables in linear regression modeling to avoid the problems of chance correlations.

We calculated 202 molecular descriptors (topological indices, TIs) for two chemical databases (a set of 139 hydrocarbons and another set of 1037 diverse chemicals). Variable cluster analysis of these TIs grouped these structures into 14 clusters for the first set and 18 clusters for the second set. Correspondences between the same TIs in the two sets reveal how and why the various classes of TIs are mutually related and provide insight into what aspects of chemical structure they are expressing.

We have used topological, topochemical and geometrical parameters in predicting: (a) normal boiling point of a set of 1023 chemicals and (b) lipophilicity (log P, octanol/water) of 219 chemicals. The results show that topological and topochemical variables can explain most of the variance in the data. The addition of geometrical parameters to the models provide marginal improvement in the model's predictive power. Among the three classes of descriptors, the topochemical indices were the most effective in predicting properties.

Two molecular similarity methods have been used to select nearest neighbors from four different sets of chemicals. One of the methods is based on the Euclidean distance of chemicals in the ten dimensional principal components space derived from 97 graph invariants. The second approach is based on the count of atom pairs common to a pair of molecules. Two probe chemicals were selected, and neighbors of each were determined by the two methods for the following four sets of molecules: (a) a combined set of octane and nonane isomers, (b) a relatively more diverse set of 382 chemicals, (c) a diverse set of 3692 chemicals, and (d) the STARLIST data base of log P consisting of 4067 structures. The results show that the measures reflect an intuitive notion of chemical similarity.

Attempts were made to develop hierarchical quantitative structure-activity relationship (QSAR) models for the dermal penetration of polycyclic aromatic hydrocarbons (PAHs) using four classes of theoretical structural parameters; viz., topostructural, topochemical, geometric, and quantum chemical descriptors; and physicochemical properties such as molecular weight (MW) and lipophilicity (log P--octanol/water). The results show that topostructural, topochemical, and geometric descriptors and molecular weight are equally effective in predicting the dermal penetration of PAHs. Quantum chemical parameters did not make any improvements in the predictive power of the QSAR models.