At the forefront of the war against cancer, researchers at the U.S. Department of Energy’s Argonne National Laboratory are leveraging one of the world’s most powerful tools, the Aurora supercomputer.
With unprecedented processing power, Aurora has been used to supercharge artificial intelligence (AI) and high performance computing (HPC), discovering effective cancer treatments, particularly for forms of disease that resist existing treatments.
The Aurora is not just another supercomputer, but an exascale system that can perform over 16 billion calculations per second.
This power allows researchers to process large datasets, simulate biological systems with extraordinary details, and train advanced AI models that cannot be processed by previous systems.
“We’ve seen a lot of experience in our business,” explains Thomas Brettin, Computational Scientist and Strategic Program Manager at Argonne.
“Part of our mission is to apply these capabilities to the epic challenges facing the nation and humanity, and cancer is one of them.”
From AI models to real-world cancer solutions
Argonne’s journey to AI-driven cancer research began almost a decade ago and was spurred by a collaboration between the Department of Energy (DOE) and the National Cancer Institute (NCI).
Early efforts have led to the development of Cancer Distributed Learning Environment (CANDLE), a scalable, deep learning framework specifically designed for the country’s growth portfolio of cancer data.
The candle paved the path of predictive models that could estimate how tumors respond to different drugs. These models laid the key foundations for Argonne’s broader strategy. It combines HPC with laboratory research to quickly track AI and lab research.
Their breakthrough came when researchers screened 500 billion small molecules in just 20 minutes using the Aurora Supercomputer. This is a task that takes several weeks on an older system. Aurora speed and scale allow for dramatic compression of the timeline of cancer drug discovery.
Pushing boundaries in predictive oncology
Based on this momentum, Argonne launched an improvement project in 2021. This initiative was established to assess and benchmark the growing ecosystem of AI models designed for cancer treatment prediction.
In partnership with Frederick National Laboratory for Cancer Research, the project has developed a standardized framework for assessing which models are best predicted across a variety of tumor types.
This work evolved once again when Argonne researchers began investigating large-scale language models (LLMS). This is an AI tool trained to generate new molecules by understanding patterns in biochemical data.
These next-generation models represent a forward leap and shift the focus from simply evaluating responses to new drug candidate designs.
Targeting “ugruggentable”
Now, the team is setting up one of the most challenging challenges in cancer research: vision. Targeted protein targets – A molecule that has long avoided treatment due to its elusive, unstable structure.
With support from the Advanced Research Projects Agency for Health (ARPA-H), Argonne is partnering with the Chicago Medical Comprehensive Cancer Center for a bold initiative called Ideal.
The goal is to find small molecules that can inhibit cancer-driven proteins that were previously thought to be untreated.
This highly interdisciplinary project combines structural biology, computational modeling, and experimental validation.
Researchers begin by identifying target proteins and, if necessary, determine the 3D structure using advanced photon sources (APS). Recently, this DOE facility has been enhanced with brighter X-ray functions.
Once the structure is mapped, the Aurora Supercomputer simulates the atomic behavior of the protein. These simulations help identify potential binding pockets. This is a small structural gap where the therapeutic molecule can attach and destroy protein functions.
From simulation to clinical testing
After computational screening, promising molecules are handed over to the laboratory team for testing. Scientists then validate AI predictions by observing how compounds interact with proteins or how they affect tumor growth in experimental models.
This simulation and real-world testing loop dramatically accelerates traditional drug discovery pipelines.
What makes this approach groundbreaking is its ability to tackle long-standing obstacles in oncology. Many of the proteins targeted by ideals are inherently disordered, meaning they have no fixed structures. This is one of the reasons they have been labelled as unugged for years.
However, with Aurora’s supercomputers and AI tools, researchers are revealing how to manipulate these elusive targets.
Why the Aurora Supercomputer is Important
The Aurora Supercomputer is more than just a powerful machine. It is the catalyst for scientific change. By significantly reducing the time required to screen for potential drugs and simulate protein behavior, Aurora helps researchers enter zero-on-one treatments, faster than ever.
This convergence of AI, HPC, and experimental science could lead a new era in oncology, providing hope for patients whose cancer has resisted traditional treatment.
Source link
