AlphaFold Protein Database Expands: 1.7M High-Confidence Complex Structures Added, Inference Speed Accelerated 100x

Nvidia, together with Google DeepMind, the European Bioinformatics Institute (EMBL-EBI), and the Steinegger Lab at Seoul National University in South Korea, has significantly expanded the AlphaFold protein structure database. The four parties jointly released the world's largest protein complex dataset, adding 1.7 million high-confidence predicted protein complexes to the searchable database, while providing approximately 30 million additional predicted structures for bulk download.

Google DeepMind's AlphaFold-Multimer model provided AI-predicted protein structures for the database, and Nvidia's compute libraries integrated into the OpenFold inference pipeline accelerated the inference process by more than 100 times. This progress will help researchers study protein interactions more efficiently and drive breakthroughs in the life sciences.