Session: Good Practice in QSAR Model Building - A Step towards Predictive Modeling of ADME/Tox Properties


Monday, March 26, 2007

4:00 PM-5:30 PM McCormick Place -- room E260
Good Practice in QSAR Model Building - A Step towards Predictive Modeling of ADME/Tox Properties

Sponsored by:	EXPOSITIONWORKSHOP
Sponsor: Strand Life Sciences, Booth 652. Instructors<:/i> Rishi R. Gupta, Eric A. Jamois, and Kas Subramanian. 4 to 5:30 PM. In-silico QSAR models still hold considerable promise in the early assessment of ADME/Tox liabilities, especially prior to synthesis. In the past few years, we have gained insight into the mechanisms governing ADME/Toxicity endpoints. Still, a healthy skepticism pervades over the accuracy of such in-silico tools, and extends to the creation, validation, and use of such models. In this workshop, we will present a set of techniques (capturing most prominent practices in drug discovery) describing how in-silico QSAR models can be created, validated, and used with increased confidence. We will describe computational methodology to estimate prediction error or even to suggest if prediction should be attempted at all. We will then investigate methods to improve the overall accuracy of predictions. Finally, we will see how custom/pre-built models can be deployed across an organization by new population of users.