Arc Institute, Stanford, & NVIDIA Unveil Evo 2, the Largest AI Model for Genomic Data

Unlike previous AI models, Evo 2 is fully open-source, with its model parameters, training code, inference code, and the OpenGenome2 dataset available to researchers.

California-based nonprofit Arc Institute, in collaboration with Stanford University and NVIDIA, has launched Evo 2, the largest publicly available artificial intelligence model for genomic data.

Evo 2 is designed to predict and design genetic sequences, including DNA, RNA, and proteins, across all domains of life. The model has been trained on nearly nine trillion nucleotides, the fundamental building blocks of DNA and RNA.

Unlike previous AI models, Evo 2 is fully open-source, with its model parameters, training code, inference code, and the OpenGenome2 dataset available to researchers.

“Deploying a model like Evo 2 is like sending a powerful new telescope out to the farthest reaches of the universe,” said Dave Burke, Arc’s Chief Technology Officer.

“We know there’s immense opportunity for exploration, but we don’t yet know what we’re going to discover.”

Applications in Healthcare & Biological Research

Evo 2 has been developed for a wide range of applications in biomolecular research. It can predict protein structures, identify novel molecules for healthcare and industrial use, and assess the functional impact of genetic mutations.

According to NVIDIA, this advancement represents a significant step forward for generative genomics.

“By advancing our understanding of these fundamental building blocks of life, we can pursue solutions in healthcare and environmental science that are unimaginable today,” said Patrick Hsu, co-founder and core investigator at Arc Institute and assistant professor of bioengineering at the University of California, Berkeley.

One of the key capabilities of Evo 2 is its ability to process genetic sequences up to one million tokens in length, allowing researchers to analyze complex relationships between genes, cell functions, and diseases.

This feature could lead to an improved understanding of gene expression and its role in medical conditions.

Evo 2 is integrated as an NVIDIA NIM microservice, enabling users to generate biological sequences with customizable settings. The model is also compatible with the NVIDIA BioNeMo Framework, allowing researchers to fine-tune it using proprietary datasets.

Brian Hie, assistant professor of chemical engineering at Stanford University and an Arc Institute innovation investigator, emphasized the potential of Evo 2 in revolutionizing biological research.

“Designing new biology has traditionally been a laborious, unpredictable, and artisanal process. With Evo 2, we make the biological design of complex systems more accessible to researchers, enabling the creation of new and beneficial advances in a fraction of the time it would previously have taken.”

AI-Powered Infrastructure

Founded in 2021 with $650 million in funding, Arc Institute supports long-term scientific research, particularly in disease areas such as cancer, immune dysfunction, and neurodegeneration.

NVIDIA contributed to Evo 2’s development by providing computing resources, including access to 2,000 NVIDIA H100 GPUs via NVIDIA DGX Cloud on AWS.

The AI model also leverages NVIDIA BioNeMo software, which features optimized microservices and BioNeMo Blueprints. NVIDIA researchers played a key role in AI scaling and optimization.

Advancing Drug Discovery & Disease Research

Evo 2’s capabilities hold significant promise in healthcare and drug discovery. Researchers can use the model to identify gene variants associated with diseases and design precise molecular interventions.

In a joint study conducted by Stanford and Arc Institute, Evo 2 demonstrated a 90% accuracy rate in predicting whether previously unrecognized mutations in BRCA1—a gene linked to breast cancer—would affect gene function.

As AI-driven genomic research continues to evolve, Evo 2 is expected to provide a transformative tool for scientists, accelerating discoveries in genetic engineering, personalized medicine, and biotechnology.

Stay tuned for more such updates on Digital Health News.