Complementarity of FTrees and Tanimoto similarity, clustering, and Bayesian learning in the search for Alpha 7 nAChR ligands

COMP 21

Wendy Sanderson, wsanders@prdbe.jnj.com1, Theo Thielemans1, Christophe Buyck1, Chris Grantham2, Anne Lesage2, Jan Willem Thuring3, and Trevor Howe1. (1) Molecular Informatics, Johnson & Johnson Pharmaceutical Research & Development, Turnhoutseweg 30, Beerse, 2340, Belgium, (2) Psychiatry, Johnson & Johnson Pharmaceutical Research & Development, Turnhoutseweg 30, Beerse, 2340, Belgium, (3) Medicinal Chemistry, Johnson & Johnson Pharmaceutical Research & Development, Turnhoutseweg 30, Beerse, 2340, Belgium
Searching for active compounds is a recurring task in the life-cycle of every drug discovery project. Without structural information of the target protein, the search for new molecules often relies on similarities and differences to known actives. There is a consequent risk of getting trapped in chemical space that has already been explored. A combination of different techniques increases chances of exploring new chemical avenues. Of particular interest to us are FTrees, which allow molecules to be described in a topology-preserving nomenclature, assigning physico-chemical properties to tree nodes.

At an early stage in the Alpha 7 nAChR project the use of FTrees (in addition to the more traditional techniques such as Tanimoto similarity on fingerprints, fingerprint clustering and Bayesian learning) led to the discovery of active molecules with hitherto unexpected scaffolds. Later in the project FTrees results were used to identify promising R-groups.

 

General Oral: Drug Discovery
8:30 AM-12:00 PM, Sunday, 10 September 2006 Moscone Center -- Room 236, Oral

Division of Computers in Chemistry

The 232nd ACS National Meeting, San Francisco, CA, September 10-14, 2006