Quantitative structure-activity relationships. 2. A mixed approach, based on Hansch and Free-Wilson analysis

Abstract

Based on the theoretical and numerical equivalence of Hansch's linear multiple regression model and the modified Free-Wilson model a mixed approach is developed. The mixed approach is a combination of both models which makes use of the advantages of each model and widens the applicability of Hansch and Free-Wilson analysis. The Free-Wilson approach now is applicable also in the case of parabolic dependence of biological activity on a particular physical property, e.g., log P or pi. A rational explanation is given for the use of dummy variables in Hansch equations and the derivation of Hansch correlations for de novo group contributions obtained from Free-Wilson analysis. Some examples illustrate the mixed approach and demonstrate its usefulness to establish biologically meaningful structure-activity relationships.

From Narcosis to Hyperspace: The History of QSAR

“…With the further progress in combinatorial chemistry it is to be expected that Free Wilson analysis may experience a revival. Both, Hansch and Free Wilson analyses, look quite different but they are inherently related [35,36]. If a Hansch model is used to calculate individual group contributions, they are very similar to the corresponding Free Wilson results.…”

“…Both models can even be combined to a ™mixed approach∫ (Eq. 9), with a Hansch relationship for substituents at a site of major structural variation and a Free Wilson part for a site with minor structural variation [35].…”

Synthesis and activity optimization of herbicidal substituted 4-aryl-1,2,4-triazole-5(1H)-thiones

“…However, to adequately predict the analogues that were only substituted ortho (Table 11,4,34-38) required use of an indicator variable IO (Kubinyi, 1976;Fujita and Ban, 1971). This variable assumed the value of 1 when the aromatic ring was solely ortho substituted (ie., 2-or 2,6-substituted or unsubstituted).…”

Structure-Activity Relationships for Dermorphin-Related Tetrapeptides

“…From a general point of view, the use of indicator variables can be considered correct when it is necessary to account for specific structural properties (c.g. confor mational) which cannot be taken into account by con-linuous LFER parameters [8], Other substituent prop erties (electronic and steric) were taken into account, but no statistically significant correlations were ob tained. This, obviously, in general, does not rule out the importance of other physicochemical factors in affecting opioid activity [e.g.…”