At this point, it is appropriate to ask the question, "are all the determinations of partition coefficients and compilation of physical chemical parameters useful only when a statistically valid QSAR model is obtained?" The answer is a firm no. One of the most useful spinoffs from the field of QSAR has been the application of experimental design to the selection of new compounds to be synthesized and tested. Let us assume that a new series of drug molecules is to be synthesized based on the following structure. The goal is to test the effect of the 16-substituents in Table 2.6 at each of the three positions on our new series. The number of possible analogs is equal to 163, or 4,096, compounds, assuming that all three positions will always be substituted with one of the substituents from Table 2.7. If hydrogen is included when a position is not substituted, there are 173, or 4,913, different combinations. The problem is to select a small number of substituents that represent the different ranges or clusters of values for lipophilicity, electronic influence, and bulk. An initial design set could include the methyl and propyl from the
Was this article helpful?