Sept 8-9 Predictive Tox Event -- "QSAR - Got to Know Its Limitati on" Presentation

Presenting Barnett International's
Predictive Toxicology
Strategies to Reduce Late Stage Drug Failures and Address Bottlenecks of
In-vitro and In-vivo Toximetabolism Using High-Throughput Methods
September 8-9, 2004 * Park Hyatt Hotel * Philadelphia, PA
http://www.barnettinternational.com/RSC_PDFUploads//BI4007WEB.pdf

This comprehensive event featuring industry, academic and agency case
studies offers strategies to reduce late stage drug failures and address
bottlenecks of in-vitro and in-vivo toximetabolism.

There will be a special presentation by Weida Tong, Director of Center for
Toxicoinformatics, National Center for Toxicological Research (NCTR), FDA
titled,

"QSAR - Got to Know Its Limitation" - A Regulatory Perspective
SAR/QSAR has been widely applied in drug discovery, lead optimization,
ADME/Tox prediction and regulatory decisions. A SAR/QSAR model will fail one
way or another while predicting unknown chemicals due to the limitation of
the training set in size and structural diversity as well as the nature of
the correlation method used. Therefore, the current challenge in SAR/QSAR is
no longer in constructing a fitted model with sound statistic; rather, it is
how to quantify a model's limitation for predicting unknown chemicals. A
novel QSAR method, Decision Forest (DF), will be presented, which overcomes
technological limits associated with most common SAR/QSAR methods and
permits easy assessment of a model's limitation. DF is an ensemble
technique, which combines the results of multiple heterogeneous but
comparable Decision Tree models to produce a consensus prediction. The
degree of random correlation, prediction confidence and domain extrapolation
can be readily estimated using an extensive cross-validation and
randomization testing. The method will be demonstrated on predicting
estrogen receptor binding from a regulatory perspective.

When registering as a member of the QSAR and Modelling Society please
request your 15% discount.

Phone: (610) 565-2622, ext. 2200
Email: customer.service*parexel.com
Received on 2004-08-18 - 15:46 GMT