InhibOx Case Study

InhibOx is the developer of Scopius, the world’s largest virtual library of modeled drug candidates. The company delivers Scopius, along with the systems to build custom versions for customers and proprietary searching technologies to its customers in the biotech, pharmaceutical, and agrochemical sectors. InhibOx also performs contract drug discovery research using these technologies.

The business model of InhibOx requires virtually unlimited computing power, and a very high level of data security. The company was exploring infrastructure options that could provide adequate power and security, but also allow the company to reduce computing costs.

Paul Davie, CEO of InhibOx, describes why they chose to work with Amazon Web Services (AWS): “The reliable availability of effectively unlimited computing power is critical to our business model, and AWS has proved faultless in delivering this, with a very high level of customer service. Secondly, AWS provides the highest levels of data security, a critical factor for us and our clients.”

InhibOx uses AWS to provide the huge computational resources needed to build Scopius-format databases, with pre-calculated 3D structures and properties. Paul Davie says, “We used Amazon Elastic Computer Cloud (Amazon EC2), Amazon Simple Storage Service (Amazon S3), and Amazon EC2 Spot Instances to construct massive compute clusters as needed. After benchmarking the most appropriate types of compute instances, we settled on cc2.8xlarge as giving the most cost-effective computational throughput."

“We developed our own proprietary software that used a wide variety of software on Ubuntu, including bash, Python, and C++. Our cloud infrastructure was built on MIT StarCluster, S3tools, and AWS Import/Export. StarCluster was a great help in replicating on the cloud the kind of computational infrastructure we normally use internally. Amazon CloudWatch proved particularly useful. Amazon S3 tools helped to store our input files and backup our computational output. AWS Import/Export was used to set up and manage the transfer of our results to our customer.”

Recently, InhibOx ran a four-day job in the U.S. Standard region, across 64 cc2.8xlarge instances. Dr Garrett Morris, InhibOx’s Research Director, explains the benefits of this operation: “We used Amazon EC2 Spot Instances for the compute nodes, which generated considerable cost savings for us. Had we attempted to construct this hardware infrastructure in-house it would have cost $192K, plus other associated costs. So the savings ran into the hundreds of thousands of dollars.”

Another benefit is the ease of launching computational clusters. Dr Morris notes, “We found that there is now a wealth of open source tools that have matured to the point where it is much easier to launch and use a computational cluster on AWS than just a year ago. Not only that, but AWS provides access to a scale of resources that permits the construction of systems that rival the performance of the top 500 supercomputers, with the availability of Amazon EC2 Spot Instances, for a tiny fraction of the cost.”

Davie also appreciates being able to depend on AWS: “I like the fact that InhibOx can rely absolutely on AWS providing us with as much computational power as we need, whenever we need it, and that it can do this so cost-effectively and securely. This has transformed our business.”

According to Davie, InhibOx will continue to use AWS for large projects. He says, “We will continue to explore the ever-expanding and already rich set of cloud technologies that AWS is releasing.”