The screening and deconvolution of Torrey Pines Institute' libraries leads to the rapid identification of lead individual compounds as well as information on structural analogs. Inherent in the deconvolution of the Torrey Pines Institute libraries is a certain amount of structure activity relationship ("SAR") data. This information alone or in conjunction with prior knowledge of the target(s) and/or additional leads is used to optimize the current leads.
When specific information is known about the target from crystallography or NMR studies, three dimensional models can be used to optimize leads. For example, leads derived from our libraries and other compound collections can be "docked" into the model targets to both identify putative binding modes as well as identify opportunities for increasing activity. As an example, in previous studies using docking models obtained with small molecules and protein kinases specific modifications to leads were identified. These structural modifications were designed to improve activity by including nitrogen atoms at specific positions in order to make hydrogen bonds with the target.
Another use of docking approaches is the virtual or in silico screening of synthetic combinatorial libraries and other compound collections. Using this approach we have identified novel protein kinase B and DNA metiltransferase inhibitors as potential therapeutic agents for the treatment of cancer. The outcome of such calculation is to predict what molecules within a library or collection are most and least likely to be active with a given receptor. Docking different combinatorial libraries that differ in the core scaffold with the same target enables the prediction of what library is the most likely to have activity. This process can be seen as a docking-based scaffold ranking and can be used to complement our traditional approach.