Dear All, We plan to release an updated set of AlphaFold structure predictions on AFDB in the coming weeks. This will update a subset of predictions affected by a temporary numerical bug (miscompilation). This temporary issue resulted in low accuracy predictions with correspondingly low pLDDT for ~4% of the total structure predictions available in the database. As these predictions had low pLDDT, we hope that this bug will not have significantly impacted any work.
This does not impact the AlphaFold model or open source code. We’ve made a full list of impacted structure predictions available at - https://ftp.ebi.ac.uk/pub/databases/alphafold/v3_affected_accessions.txt. This set of structures were part of the recent 200M expansion and didn’t impact any of the previous releases (model proteomes released in V1 and V2, or UniProtKB/Swiss-Prot released in V2). The new release (V4) will include: Updated coordinates for affected structures (~4% of total structures). You can still access all old coordinates as V3 files, and easily compare V3 and V4 coordinates. Minor metadata changes in the mmCIF files for the rest of the structures (these files will be released as V4) We’ll be in touch after releasing updated structures soon. As always, we value community engagement and feedback and would like to thank Igor Tolstoy (NCBI) for surfacing the large pLDDT differences in highly similar sequences that alerted us to the issue. Please get in touch via alphaf...@deepmind.com <mailto:alphaf...@deepmind.com> if you have any feedback or require support with this change. Best Wishes, Sameer Velankar (EMBL-EBI) and Tim Green (DeepMind) ######################################################################## To unsubscribe from the CCP4BB list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1 This message was issued to members of www.jiscmail.ac.uk/CCP4BB, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/
