July 10, 2014

Advancing Drug Discovery with HPC Cloud

Tiffany Trader
Cycle one drug cost

In order to develop a new drug the average pharmaceutical company spends about one billion dollars over ten years. This is an untenable situation for those with serious illnesses who don’t have the luxury of time. That’s why high-performance computing, which has been shown to dramatically accelerate the pace of discovery, is so critical to the domain of bio-medicine.

Novartis Institutes for Biomedical Research is one of the bioscience companies who understands the advantages of using HPC in the fight against diseases like cancer. In a presentation at the 2014 AWS Summit in New York today, Novartis detailed how it has used the Amazon cloud to identify promising drug compounds.

The company relied on AWS EC2 and Cycle Computing software to create and access a $44 million supercomputer for just $5,000. In only 11 hours, the Novartis researchers had three potential cancer-fighting compounds to run further trials on.

The entire collaboration was facilitated by Cycle Computing as described by Cycle’s Brad Rutledge in a recent blog entry.

Cycle Computing How

To better assess specific target proteins associated with certain cancers, Novartis researchers used Cycle’s cloud software in tandem with Amazon’s EC2 cloud, enabling them to spin up a massive supercomputer in record time and with little upfront cost. In terms of specific software, Novartis relied on CycleSever for workload placement, CycleCloud to orchestrate HPC environments across the Amazon infrastructure, Chef 11 Server for continuous instance configuration, and HTCondor for job scheduling.

In 11 hours, the team completed the equivalent of 39 drug design years for a cost of $4,232. During that run, Novartis screened 10 million compounds and identified three promising candidates, which have been promoted to the next stages of drug design, bringing the company that much closer to a life-saving cancer drug.