Could computers predict the next big Alzheimer's drug?

Scientists are developing a computational process for designing drugs that could trump chemists in speed and efficiency. By doing so, the technology could reduce some of the cost and risk that have plagued drug discovery and made the expensive hunt for new therapies an unattractive endeavor for Big Pharma.

At the University of Dundee, researchers engineered the system to mimic the iterative approach of medicinal chemists as they craft compounds to deliver the safety and efficacy profiles needed to combat diseases. They used the structure of the Alzheimer's drug donepezil to test the system, which "evolved" the structure of the treatment automatically and predicted its profile for a variety of drug targets. The group highlighted that 75% of the predictions were confirmed.

According to the Scottish university, the system could aid in the design of drugs for complex diseases in fields such as neuroscience, cancer and infectious diseases.

Testing the system with an approved drug was just a starting point for proving the concept works. The researchers see a commercial opportunity for the "automated adaptive design" approach, and a startup called Ex Scientia (Latin for "from knowledge") was spun off from the project. Indeed, Big Pharma could see big value in shaving time and expense from the years and millions of dollars typically required for drug discovery to bear fruit.

"This proof of concept shows that we could make significant advances in discovering and designing complex drugs, which could lead to improvements in safety and efficacy," Andrew Hopkins, chief of medicinal informatics at Dundee, stated in a release, "while also potentially reducing the cost of drug discovery, which is a high-risk and expensive process."

The project has benefited from the surge in the amount of data on drug design from sources such as The Wellcome Trust-supported ChEMBL database of small molecule structures. In October, GlaxoSmithKline ($GSK) unveiled new efforts to open large amounts of R&D data to scientists for further research, including a library of about 200 compounds that could provide new treatments for tuberculosis.

- here's the release
- see the Nature abstract
- and an article from the BBC