Assessment of additive/nonadditive effects in structure-activity relationships: implications for iterative drug design.

Free-Wilson (FW) analysis is common practice in medicinal chemistry and is based on the assumption that the contributions to activity made by substituents at different substitution positions are additive. We analyze eight near complete combinatorial libraries assayed on several different biological response(s) (GPCR, ion channel, kinase and P450 targets) and show that only half-exhibit clear additive behavior, which leads us to question the concept of additivity that is widely taken for granted in drug discovery. Next, we report a series of retrospective experiments in which subsets are extracted from the libraries for FW analysis to determine the minimum attributes (size, distribution of substituents, and activity range) necessary to reach the same conclusion about additive/nonadditive effects. These attributes can provide guidelines on when it is appropriate to apply FW analysis as well as for library design, and they also have important implications for further steps in iterative drug design.