Abstract:
The broad field of quantitative structure-property relationships (QSAR) allows researchers in many areas to use such dependencies to correlate properties with particular features of the compounds examined. After this information is gleaned, the next step is the use the results on the relative importance of some features to predict information about compounds that have not been examined in this way. The hope is, of course, that better compounds can be developed in this manner possessing optimal desired properties. Unfortunately, the limiting step in the procedure is how to best quantify the particular features of a compound's structure. In this presentation, I will discuss successful strides made on this problem through the introduction of a molecular descriptor called Signature, which quantifies a molecule's 2D structure. I will highlight a variety of successes we have had in solving inverse design problems with Signature (ICAM-1, HFEs, COX-2) as well as where Signature has found utility in similarity searches and other arenas.