InkSpot. Science. On Demand

Open QSAR

---

• Author: David Leahy
• November 9, 2009
• Category: Uncategorized, biologics

Open QSAR is a new site for the development and assessment of comparative QSAR modelling methodology (meta QSAR) as well being available for QSAR model building and multi-property prediction. The site uses an automated QSAR building and prediction software system that has been developed over the last few years and which is known as the Discovery Bus.

At this stage the site displays 15,000 QSAR models as well as validation data for around 100 properties extracted from the WOMBAT database of Prof. Tudor Oprea of the University of New Mexico. We evaluate the models against criteria outlined by Tropsha et al in their paper “The Importance of Being Earnest”, QSAR Comb. Sci. 22 (2003), 69.
We will shortly be extending this to include the remainder of the WOMBAT database as well as the recently released ChEMBL database comprising 2 million data points for over 5,000 targets. This study, which we are calling the “Mother of All QSAR’s” is supported by Microsoft Research who are providing the Azure cloud computing nodes that we need to reduce the time required from several years to a few weeks. We expect this study to create a comprehensive QSAR model library that can be used for multi-property predictions in drug discovery. We are now testing its use for predictions and will make that available for use shortly. It is also our intention to offer the site for training models.

This is the first of a series of posts on the technical capabilities of this new resource and announcements as it develops. We would be grateful for feedback and questions.

• 3 Responses
• Leave a comment...
• Comment Feeds