In a landmark advancement for computational biology, NVIDIA, in collaboration with the Arc Institute and Stanford University, has launched Evo 2, a revolutionary AI model designed to transform biomolecular research. Evo 2 is heralded as the largest publicly available AI model for genomic data, capable of decoding and designing the genetic code—DNA, RNA, and proteins—across all domains of life, including plants, animals, and bacteria. This cutting-edge development, detailed on NVIDIA’s blog, promises to accelerate discoveries in healthcare, agriculture, and environmental science, marking a significant leap forward in the intersection of artificial intelligence and biology. This powerful biomolecular AI model was trained on a dataset of nearly 9 trillion nucleotides and is now accessible to scientists.

AI Authority Trend: LatchBio Simplifies GPU-Powered Multi-Omics Tools Access with NVIDIA

Evo 2 is now accessible to global developers via the NVIDIA BioNeMo platform, including as an NVIDIA NIM microservice for streamlined and secure AI deployment. The model’s capabilities are vast: it can predict protein structures and functions from genetic sequences, identify novel molecules for medical and industrial use, and assess how gene mutations impact biological processes.

The model’s development was supercharged by NVIDIA’s contribution of 2,000 H100 GPUs through DGX Cloud on AWS, providing Arc Institute researchers with the computational power to handle massive datasets and accelerate discoveries. NVIDIA’s BioNeMo Framework, an open-source suite of tools, allows developers to fine-tune Evo 2 with proprietary data, while the NIM microservice enables users to generate tailored biological sequences by adjusting model parameters.

Founded in 2021 with $650 million from its donors, the Arc Institute fosters innovative, long-term research by offering scientists eight-year renewable funding terms, state-of-the-art labs, and partnerships with leading institutions like Stanford, UC Berkeley, and UC San Francisco. This unique environment, combined with NVIDIA’s cutting-edge technology, empowers researchers to tackle complex challenges in fields like cancer, immune dysfunction, and neurodegeneration.

Evo 2’s novel architecture sets it apart, capable of processing genetic sequences up to 1 million tokens long. This extended range offers scientists a deeper look into the genome, potentially revealing connections between distant genetic regions and their roles in cell function, gene expression, and disease. In healthcare, Evo 2 could transform drug discovery by pinpointing disease-linked gene variants and designing targeted therapies. For instance, tests by Stanford and Arc Institute researchers showed Evo 2 could predict with 90% accuracy whether previously unknown BRCA1 mutations—linked to breast cancer—would disrupt gene function.

Beyond medicine, Evo 2’s applications span multiple domains. In agriculture, it could enhance food security by enabling the development of climate-resilient, nutrient-rich crops. In environmental science, the model might aid in designing biofuels or engineering proteins to degrade pollutants like oil or plastic. Trained on genetic data from diverse species—plants, animals, and bacteria—Evo 2 is a versatile tool for advancing biotechnology and materials science.

AI Authority Trend: Stelia Named Gold Sponsor for NVIDIA GTC, Showcasing Hyperband and AI Scalability Discussion

FAQs

1. What is Evo 2?

Evo 2 is a powerful, publicly available AI foundation model designed to understand the genetic code across all domains of life. Trained on a massive dataset of nearly 9 trillion nucleotides, it’s the largest AI model for genomic data to date. It can be used for biomolecular research applications like predicting protein function, identifying novel molecules, and evaluating the impact of gene mutations.

2. What are some potential applications of Evo 2?

Evo 2 has wide-ranging applications across biomolecular sciences. In healthcare and drug discovery, it can help identify gene variants linked to diseases and design targeted treatments. In agriculture, it can assist in developing climate-resilient and nutrient-dense crops. It can also be used in materials science to design biofuels or engineer proteins for various purposes, such as breaking down oil or plastic.

3. Who developed Evo 2, and what resources were used?

Evo 2 was developed in a collaboration between the Arc Institute and Stanford University. NVIDIA accelerated the project by providing access to 2,000 NVIDIA H100 GPUs via NVIDIA DGX Cloud on AWS. NVIDIA also provided expertise in AI scaling and optimization.

Conclusion

Evo 2 represents a significant leap forward in generative genomics, offering unprecedented capabilities for understanding and manipulating the building blocks of life. Its ability to process vast genetic sequences and predict the effects of mutations opens up a world of possibilities in healthcare, agriculture, and other scientific fields. As scientists begin to explore the potential of Evo 2, world can anticipate groundbreaking discoveries and innovative solutions to some of the world’s most pressing challenges.

AI Authority Trend: Eficode Bridges Software Development and GenAI with NVIDIA AI Enterprise

To share your insights, please write to us at news@intentamplify.com