Significant research on which metrics are useful in analyzing molecular diversity has been recently reported. The use of 2D fingerprints based on chemical fragments or atom types and bond paths have emerged as a metric which is computationally fast as well as having reasonable neighborhood properties. Clustering based on these 2D fingerprints has therefore been used to help select reagents for combinatorial chemistry as well as compounds to acquire from external sources. While clusters based on 2D fingerprints are useful for grouping compounds in a database they do have some limitations. Here we present an alternative method for analyzing a molecular graph based on ring templates as well as user-defind templates. Results and examples of how these templates can be used in molecular diversity analysis will be presented.

I will also talk about our application of several Daylight toolkits as they apply to the diversity analysis described above, and will be making a demo during the presentation.