NVIDIA and partners release viral protein dataset for pandemic research

NVIDIA has joined Google DeepMind, EMBL-EBI and other global research organizations to publish predicted 3D structures for protein complexes from more than 2,800 viruses in the AlphaFold Database. The dataset was generated with AlphaFold2, optimized by NVIDIA BioNeMo Inference Runtime, allowing researchers to scale predictions across thousands of viral proteomes and study groups of interacting proteins encoded by each virus.

The effort is intended to build scientific readiness before another major outbreak. The Center for Global Development estimates a roughly 50% chance of a pandemic as severe as COVID-19 by 2050, and project collaborators argue that advance structural knowledge could help scientists move faster when unfamiliar viruses emerge. About 30% of the newly added protein interactions are described as completely new to science, with shapes not previously documented in the Protein Data Bank.

NVIDIA is also releasing the BioNeMo Structure Prediction Pipeline, the GPU-accelerated workflow used to create the dataset, so scientists can move from protein sequence to predicted 3D structure for their own targets. The AlphaFold Database now contains more than 260 million protein and protein complex predictions, with the new viral data labeled by confidence and made available for researchers studying diagnostics, treatments and vaccines.