Relay Therapeutics Uses AWS to Accelerate Drug Discovery

Relay Therapeutics is a precision medicine company transforming the drug discovery process by leveraging unparalleled insights into protein motion. Prior to testing promising compounds in the lab, scientists have to consider a molecular universe of available starting points numbering close to 10 billion compounds. They need to filter this extensive set down to the 100–200 compounds most likely to bind to the biological target.

By analyzing more compounds, scientists increase the chances they will find the right molecules to test in the lab. In a typical on-premises data center, with thousands of CPUs, the analysis of a billion compounds could take months. Deploying sufficient CPUs in an on-premises data center would also be cost-prohibitive, particularly due to the “bursty” nature of the analyses.

Typically, in traditional IT environments, pharmaceutical companies virtually screen a few million compounds at a time. Relay Therapeutics was determined to scale that number into the billions and turned to Amazon Web Services (AWS) to solve the challenge. “The major factor in selecting AWS over other cloud providers is the support we received from the start,” says Pat Walters, senior vice president of computation at Relay Therapeutics. “And it has continued to help us make our processes work more efficiently.”

On AWS, the company also simplified virtual screening so scientists can use open source scripts to kick off analysis on AWS Batch. The scientists then rapidly analyze the data by taking advantage of Amazon Athena, a serverless query service with no infrastructure to manage. 

Pierce estimates that Amazon EC2 Spot Instances reduce compute costs by 50 percent compared to conducting virtual screening on premises. AWS and Relay Therapeutics also built parameter checks into the process to keep analysis costs from exceeding the budgeted amount. “We get alerted if a job will go beyond a set expense threshold,” Walters explains. “That tells us a parameter is off so we can terminate the job or make an adjustment on the fly.”

Since deploying the AWS high-performance computing solution, Relay Therapeutics has run multiple screens of five billion compounds. Because of the scalability offered by AWS, scientists can run the screens on multiple snapshots of the same moving protein target.  

A few years ago, the Relay Therapeutics team did not think it was possible to run virtual screening at the scale the company has now achieved, with scientists analyzing tables with a billion rows. “Even sorting a table with billions of rows is not a trivial exercise,” Walters emphasizes. “By using AWS technologies, we can deal with all that information efficiently, which helps us strive toward our ultimate goal—getting medicines to patients faster than we previously thought possible.”