[3dem] AlphaFold DB - free and open access to (millions of) protein structure predictions

[Gerard Kleywegt]

Dear colleagues and friends,

I would like to share some exciting news that has been announced today. 
EMBL-EBI and DeepMind have co-developed the AlphaFold Protein Structure 
Database (AlphaFold DB; https://urldefense.proofpoint.com/v2/url?u=https-3A__alphafold.ebi.ac.uk&d=DwIDaQ&c=-35OiAkTchMrZOngvJPOeA&r=L7-zyQ-04fFCMRqzLIOnx7H0exGZHwIQe_wMPuY600I&m=UiSo5wRZXOje9l7WNIWCSdWmWxqpRx7RA2w7mNSgWs4&s=zNOkVsJTkm_K0bEkqyRgZoq-jbFoUFOJzMV1c8DY_zQ&e= ), a joint project to openly 
and freely share millions of AlphaFold protein-structure predictions with the 
scientific community. The database launched officially at 4 pm UK time on 22 
July. Today’s release contains approximately 365,000 structures (covering over 
20 reference proteomes), which will increase to an estimated 130 million (!) 3D 
models in the coming months (covering all UniProt sequence clusters with up to 
90% mutual sequence identity, i.e. UniRef90). A Nature paper describing  the 
predictions for the human proteome and mentioning the new AlphaFold DB resource 
was made public today: https://urldefense.proofpoint.com/v2/url?u=https-3A__www.nature.com_articles_s41586-2D021-2D03828-2D1&d=DwIDaQ&c=-35OiAkTchMrZOngvJPOeA&r=L7-zyQ-04fFCMRqzLIOnx7H0exGZHwIQe_wMPuY600I&m=UiSo5wRZXOje9l7WNIWCSdWmWxqpRx7RA2w7mNSgWs4&s=i-sJ9Do4kVHdx_wWBmyUAv1yDgi-sUgiOzE6Tome-dk&e= 

The AlphaFold DB resource has been the work, carried out over a period of about 
three months, of scientists, IT specialists, web designers, comms people etc. 
at both EMBL-EBI and DeepMind, with the PDBe-KB team (https://urldefense.proofpoint.com/v2/url?u=https-3A__pdbe-2Dkb.org_&d=DwIDaQ&c=-35OiAkTchMrZOngvJPOeA&r=L7-zyQ-04fFCMRqzLIOnx7H0exGZHwIQe_wMPuY600I&m=UiSo5wRZXOje9l7WNIWCSdWmWxqpRx7RA2w7mNSgWs4&s=IV84A8RzX827l3yVdd_dpZDe-b1-Sk0x1Al8cTRPi00&e= ), 
led by Sameer Velankar, playing a major role.

Given the accuracy demonstrated by AlphaFold models to date, this resource is 
likely to have a major impact not only on structural biology but on many fields 
of science and biotechnology. Soon, for the first time in history, for every 
protein sequence known to science, there will be either an experimental 
structure in the PDB, or a 3D model in AlphaFold DB, or a structure for a 
protein within “homology-modelling distance” of a target protein. The source 
code of AlphaFold has been made open as well, so predictions for completely new 
and non-natural (designed) sequences can be generated by anybody who wants to.

Speaking from experience, it may take some time to wrap your head around the 
sheer scale of the new resource and to ponder its potential impact on science. 
A small group of leading structural biologists within EMBL have produced a 
briefing document 
(https://urldefense.proofpoint.com/v2/url?u=https-3A__www.embl.org_news_science_alphafold-2Dpotential-2Dimpacts_&d=DwIDaQ&c=-35OiAkTchMrZOngvJPOeA&r=L7-zyQ-04fFCMRqzLIOnx7H0exGZHwIQe_wMPuY600I&m=UiSo5wRZXOje9l7WNIWCSdWmWxqpRx7RA2w7mNSgWs4&s=rcaNMZKutgqun36JA6xEsC8pZmW7z6wX1x4CnazA9Hk&e= ) that describes 
the technical achievements, the current limitations of AlphaFold and some of 
the potential applications and opportunities for new research in a number of 
(mainly structure-related) fields.

I for one am immensely excited and I invite you all to check out the new 
resource.

Best wishes,

--Gerard

---
Gerard J. Kleywegt, EMBL-EBI, Hinxton, UK
Head of Molecular and  Cellular Structure
gerard at ebi.ac.uk pdbe.org emdb-empiar.org
PA: Roisin Dunlop    pdbe_admin at ebi.ac.uk