Case Study: Atomwise accelerates drug discovery with Weka and AWS

How Atomwise Accelerated and Innovated Drug Discovery and Time to Market with WEKA and AWS

Atomwise, a pharmaceutical research company using AI for structure-based drug discovery, needed a way to move petabytes of unstructured data into its training pipelines more efficiently. As its workloads grew to tens of millions of files and dozens of developers, the company’s biggest challenge became I/O bottlenecks in its storage system, which slowed model training and extended compute times.

Weka, working with AWS, implemented Weka clusters for Atomwise’s Amazon EKS environment using Amazon EC2 i3en instances and shared access to Amazon S3 and Amazon EFS data. The result was dramatically faster performance: 1GB file create-and-copy times improved 40x, small file access improved 3x, and model training times dropped by up to 2x. With Weka, Atomwise could run experiments in less than a week instead of months and scale compute on demand.

Atomwise

Jon Sorenson

PhD, VP of Technology Development