AlphaFold’s protein structure predictions now available to explore

Protein structures representing the data obtained via AlphaFold. Source image: AlphaFold. Design credit: Karen Arnott/EMBL-EBI
23 July 2021

DeepMind and EMBL’s European Bioinformatics Institute (EMBL-EBI) have launched the AlphaFold Database (AlphaFold DB), a joint project to openly and freely share millions of AlphaFold protein structure predictions with the scientific community.

 

The scientific research community has long understood the importance of protein structure in progressing our understanding of biological processes. The PDB archive now contains over 180,000 experimentally determined structures and here at PDBe we are proud to manage this vital archive alongside our partners in the Worldwide Protein Data Bank (wwPDB).

 

The complexities of structural biology techniques mean that, despite the immense effort of researchers over the past decades, the number of proteins with known structure is still just a tiny fraction of those known to exist in biology. The availability of accurate protein structure predictions in AlphaFold DB can provide new and essential insights into the function and interactions of an even larger collection of proteins. These can also support structural biologists in providing accurate starting models to support their own structure determination efforts.

 

The first AlphaFold DB release contains approximately 365,000 protein structures, covering the proteomes of 22 species, including human, mouse, E. coli and more. The ambition is to grow the database to around 130 million proteins in the coming months - that’s around 700 times the number in the PDB! This would mean that for almost every known protein sequence there will either be a 3D model available from PDB or AlphaFold DB, or its sequence will be within homology-modelling distance from such a model. For completely new sequences, the AlphaFold software can be used to generate models.

 

This development could bring about a paradigm shift in molecular biology and many other fields that benefit from structural insights. The vast database of structural predictions can be used in basic research as well as a huge range of applications, from developing new drugs and therapies to designing new crops that resist climate change or developing enzymes that degrade plastic.

 

AlphaFold is an AI system developed by DeepMind that predicts, with state-of-the-art accuracy, a protein’s 3D structure from its amino-acid sequence. In November 2020, AlphaFold was recognised by the assessors of CASP14 as the best-performing method to date for predicting protein structure. For the first time ever, predictions from an AI method – AlphaFold – were consistently similar to the experimentally determined structures of the proteins included in this round of CASP.

 

Built on DeepMind’s AI expertise and EMBL-EBI’s decades of experience in curating and sharing the world’s biological data, the AlphaFold Database complements existing experimental and computational methods.

Visit the AlphaFold Database to see if the predicted 3D structure of your protein of interest has already been added.