Computational analysis of activity and selectivity cliffs.

The exploration of structure-activity relationships (SARs) is a major challenge in medicinal chemistry and usually focuses on compound potency for individual targets. However, selectivity of small molecules that are active against related targets is another critical parameter in chemical lead optimization. Here, an integrative approach for the systematic analysis of SARs and structure-selectivity relationships (SSRs) of small molecules is presented. The computational methodology is described and a cathepsin inhibitor set is used to discuss key aspects of the analysis. Combining a numerical scoring scheme and graphical visualization of molecular networks, the approach enables the identification of different local SAR and SSR environments. Comparative analysis of these environments reveals variable relationships between molecular structure, potency, and selectivity. Furthermore, key compounds are identified that are involved in the formation of activity and/or selectivity cliffs and often display structural features that determine compound selectivity.