Predictions from Hansch analysis

A second problem for Hansch analysis and any other predictive method is combinatorial chemistry. As methods in combinatorial chemistry continue to be refined and advanced, directed combinatorial libraries constructed around the scaffold of a lead have become a more standard practice. The availability of hundreds of lead analogues greatly diminishes the potential contribution from a Hansch equation. Why predict a structure s activity when one can make a library of essentially all interesting analogues, screen the library, and know the activity for certain ... 

Hansch analysis Hansch analysis is a common quantitative structure-activity relationship approach in which a Hansch equation predicting biological activity is constructed. The equation arises from a multiple linear regression analysis of both observed biological activities and various molecular property parameters (Hammett, Hansch, and Taft parameters). 

Table I shows the various ways in which w and log P have been applied to Hansch analysis. In the first equation, log P refers to the complete molecule, and an optimum value is predicted from random walk theory when drug transport is rate determining. If this is applied as a model equation to complex molecules where additivity of tt constants does not apply, log P must be measured, or large deviations will occur.

Often predictive ability is considered to be a criterion for the relevance of quantitative structure-activity analyses. While this is an obviously reasonable demand, it should be realized that the main purpose of Hansch analysis is a better understanding, not prediction. New hypotheses can be established from quantitative analyses, which are proven or disproven by synthesis and testing of new analogs. If the predicted values are close to the experimental ones, the model can be accepted. 

The most appropriate and widely used method for extracting information from large data sets is QSAR and its relatives, quantitative structure-property relationships (QSPR) for property modeling, and quantitative structure-toxicity relationships (QSTR) for toxicity modeling. QSAR is a simple, well validated, computationally efficient method of modeling first developed by Hansch and Fujita several decades ago (30). QSAR has proven to be very effective for discovery and optimization of drug leads as well as prediction of physical properties, toxicity, and several other important parameters. QSAR is capable of accounting for some transport and metabolic (ADMET) processes and is suitable for analysis of in vivo data. 

---