NVIDIA, Google DeepMind, and the European Molecular Biology Laboratory’s European Bioinformatics Institute (EMBL-EBI) have released predicted 3D structures for the protein complexes of more than 2,800 viruses. The dataset is now available to the scientific community through the AlphaFold Database.
PartnershipsNvidiaGoogle DeepMindEuropean Bioinformatics InstituteAlphaFold DatabaseAlphaFold2
NVIDIA and DeepMind Release 3D Protein Complex Structures for Over 2,800 Viruses
The predicted structures were generated using AlphaFold2, optimized with the NVIDIA BioNeMo Inference Runtime, which allowed the team to scale inference to thousands of viral proteomes. This process enables the prediction of protein complexes, which are groups of interacting proteins, at a scale and speed much faster than traditional experimental methods like X-ray crystallography.
Approximately 30% of the protein interactions added to the database are entirely new to science, representing shapes that have not been previously documented in the Protein Data Bank. The release aims to provide scientists with foundational biological data to better understand viral mechanisms and accelerate the development of vaccines and antiviral treatments for future outbreaks.
In addition to the dataset, NVIDIA has released the BioNeMo Structure Prediction Pipeline, a GPU-accelerated workflow that allows researchers to predict 3D structures from protein sequences for their own targets.
Sources
- How Open Science Can Help Researchers Prepare for the Next Pandemic （NVIDIA Generative AI, 2026-09-24）
- AlphaFold Database