The Pentagon's newest toy isn't a fancy warfighting machine - it's a combined supercomputer and rapid response laboratory (RRL) dedicated to beefing up the US's biodefenses.

Located at Lawrence Livermore National Laboratory (LLNL) in California, the new machine, built in cooperation with the National Nuclear Security Agency, will feature the same architecture as LLNL's upcoming El Capitan exascale supercomputer, with its AMD MI300A APUs.

Specifics of the system's hardware, and whether it will use existing drug discovery algorithms or government-developed ones, weren't shared.

It will be used to advance both military and civilian defenses against biological threats using large-scale simulations, AI modeling, classification of threats, and, with the combination of a new RRL, to speed the development of medical countermeasures. 

"Some of [the] countermeasures that we're going to develop in this unique environment are going to be extremely important, because we're talking about taking solutions down to days, if not hours," said Darryl Colvin, joint program executive officer for Chemical, Biological, Radiological, and Nuclear Defense. "This computing power is going to be something that we're going to use on a regular basis."

And let's be realistic - AI drug development systems can also be modified to create deadly bioweapons, not that we're saying the Pentagon would use its new supercomputer to that end.

For those unfamiliar with the concept of biodefense, it's a collection of measures used to counter both natural and man-made biological threats to military assets, as well as civilians and environmental resources like water and food, impacts to any of which can have a negative effect on warfighters' abilities. 

Because of the wide-reaching impact of biological threats, use of the new supercomputer will also be available to other US government agencies, international allies, academia, and industry. 

The RRL, on the other hand, is being stood up just "a short walk away from the computing facility" to complement the Department of Defense's Chemical and Biological Defense Program's Generative Unconstrained Intelligent Drug Engineering (GUIDE) program by marrying its existing capabilities with advanced supercomputing.

GUIDE focuses on developing medical countermeasures "by leveraging machine learning-backed antibody design, experimental data, structural biology, bioinformatic modeling, and molecular simulations," LLNL said. The new supercomputer "will allow DoD and the NNSA Tri-Labs [LLNL, Sandia, and Los Alamos] to perform rapid and iterative testing of computationally designed vaccines and antibody drugs."

That lab is automated too. According to LLNL, the RRL includes molecular characterization rooms that contain automated robots and machinery that can simultaneously re-engineer proteins and countermeasure candidates to accelerate the initial phases of drug discovery, which involves identifying disease targets.

Jim Brase, LLNL's deputy associate director for computing, said that having a lab connected to a supercomputing facility dedicated to improving biodefense has "transformative potential" for biological threat detection and response.

"This isn't about one project, or one particular biothreat or chemical agent, this is about building this broad national capability, so that no matter what comes up or what we see coming down the road, we can rapidly assess and respond to it," Brase said. 

It's not clear if non-DoD partners will also have access to the RRL, or if the supercomputer and lab are up and running yet. We've reached out to LLNL and the DoD to learn more. ®

https://www.theregister.com//2024/08/16/dod_drug_discovery_supercomputer/