Seemingly every supercomputer in the world is allied in the fight against the coronavirus pandemic – but not many of them are fresh out of the box. Cerebras Systems has announced that its brand new CS-1 AI supercomputer, which was deployed for the first time less than six months ago at Argonne National Laboratory, is now working to identify therapeutics for COVID-19.
For Cerebras, a team of AI-focused engineers, the Argonne deployment was a landmark. The CS-1, they claimed, was the fastest AI supercomputer in existence, including the first deployment of Cerebras’ “Wafer Scale Engine” (WSE) – itself the largest chip ever built, with 1.2 trillion transistors and nearly 72 square inches of silicon.
Argonne was its first customer for the CS-1, and the system was successfully deployed in November 2019 as the first step of a broader partnership between Cerebras and the U.S. Department of Energy (DOE). At the time, Andrew Feldman (Cerebras’ founder and CEO) highlighted how Argonne was using the CS-1 to “better understand everything from cancer drug interactions to the properties of black holes.”
A different challenge, of course, loomed on the horizon.
Many of the most powerful supercomputers are crunching COVID-19 drugs by brute force, simulating key proteins of the virus in great detail and churning through billions of molecules in the hopes of identifying those with the best chance of binding to those proteins, disabling them and rendering the virus impotent. The molecules with the best “docking scores” then get moved on to further simulations and real-world testing.
For Argonne and CS-1, brute force was not the name of the game. Instead, Argonne applied the CS-1’s AI capabilities to train machine learning models to churn through the lab’s massive molecular datasets (comprising existing FDA-approved drugs) and predict which of those molecules would have the best docking scores. The result, according to Cerebras: “hundreds of times” faster turnaround on the datasets at a fraction of the computational cost.
The first iteration of the CS-1’s battle against COVID-19 was completed over the last few weeks. Now, Argonne and Cerebras are working on a new ML process for CS-1 that would treat the process as a computer vision problem, representing viral proteins and drug molecules as images rather than numbers.
Using the output from the CS-1’s molecular analyses, Argonne is building in-silico models to hone in on the interactions between COVID-19 and the most promising drug molecules. Any molecules that clear that stage of evaluation will then be sent to the “wet lab” for real-world testing.
“In the war against COVID-19 and other novel viruses,” Cerebras stated, “the power of AI supercomputers promises to build more robust workflows, accelerate research and development of deep learning models and greatly advance the future of disease research.”
Header image: The CS-1. Image courtesy of Argonne National Laboratory.