SARtree: A new method for analyzing and visualizing the results from virtual and experimental screening of large complex chemical libraries


Donovan N. Chin, Anuj Patel, R. Aldrin Denny, and Juswinder Singh. Computational Drug Design, BiogenIdec, 14 Cambridge Center, Cambridge, MA 02142
This talk will describe a newly developed method at BiogenIdec for analyzing and visualizing the results from virtual screening of large complex chemical libraries. The method involves two unique steps. First, an algorithm to recursively partition and keep track of chemical libraries into core substructures, attached common sub-cores and unique chemical fragments. Second, an interactive graphical "tree" that permits the analyst to visualize the connection and relationships between the common and unique chemical groups from step one. Property information--e.g. virtual screening scoring, biological assay data, or both--can be overlaid on the graphical tree to allow rapid identification of chemical fragment hotspots for a given dataset. We will highlight the utility of SARtree on several chemical datasets including CDK2 kinase from virtual screening. Finally, we will demonstrate that SARtree is a powerful new way of quickly identifying chemical structure-variation and -property relationships hidden in chemical datasets. SARtree is therefore emerging as a useful tool for analyzing and understanding the output from virtual screening, and as a new way of tracking and visualizing (virtual) chemical libraries.

Advances in Virtual High-Throughput Screening
8:30 AM-11:30 AM, Thursday, August 26, 2004 Pennsylvania Convention Center -- 110A&B, Oral

8:00 PM-10:00 PM, Monday, August 23, 2004 Pennsylvania Convention Center -- Hall D, Sci-Mix

Division of Chemical Information

The 228th ACS National Meeting, in Philadelphia, PA, August 22-26, 2004