Ranking Chemical Compounds for Drug Discovery
[Summary & Contributions] | [Relevant Publications]
Summary and Contributions
The cost of developing a new drug today is estimated to be over $1 billion. A large part of this cost is the result of failed molecules: chemical compounds that appear to be promising drug candidates during initial stages of screening, but after several rounds of expensive preclinical and clinical testing, turn out to be unsuitable for further development. With chemical libraries today containing millions of structures for screening, there is an increasing need for computational methods that can help alleviate some of these challenges. In particular, computational tools that can rank chemical structures according to their chances of clinical success can be invaluable in prioritizing compounds for screening: such tools can be used to focus expensive biological testing on a small set of highly ranked, more promising candidates, leading to potentially huge savings in time and costs. We demonstrated the use of ranking methods in machine learning for ranking drug candidates. This work was featured as a spotlight on MIT News and was also covered in the press.
Relevant Publications
- Shivani Agarwal, Deepak Dugar and Shiladitya Sengupta.
Ranking chemical structures for drug discovery: A new machine learning approach.
Journal of Chemical Information and Modeling, 50(5):716-731, 2010.
[paper] [email me for a copy if you don’t have access]
Featured as an MIT News spotlight.
Also featured in HPCwire, HealthCanal, PhysOrg, Science News, US News & World Report.
Shivani Agarwal 