Maxim Tsenkov

Yes, only AF2, not AF3. Unchanged models would be the same as 2021, since the settings used here are identical to every release we've had before, and MSAs cached from previous runs for unchanged sequences. Also the sequence databases also remain the same. So the entire db is consistent throughout.

Sorry to keep you waiting Max! We got overwhelmed with questions on this release. So we wrote another post to clarify some of these queries here – bsky.app/profile/mits...

You probably heard about the last major release at #AlphaFold Database two weeks ago. We hope this new post will cover any further questions about the details of the release. If you have any other questions, please do reply here or email us your queries – afdbhelp@ebi.ac.uk

Not yet, but we are working on it! We wanted to prioritise downloads from individual entry pages, before we expand towards bulk downloads. BRB.

The latest #AlphaFold DB update is out!

We synchronised with UniProt release 2025_03, predicted protein isoforms across all species, and provide MSAs used to generate each prediction.

Proud of the team at the PDBe for the hard work that went into this. Grateful to play my part in it!

Likewise Nico! Looking forward to more developments in the near future!

This is just the start! I’ll be sharing more on this new feature on AlphaFold DB, cool use-cases, and various other updates.

Have you got an interesting domain example? Did you find a case where TED annotations shine (or break) in our new feature?

Let me know! Follow for more updates! 🔜

🧵(6/6)

And, of course, TED wouldn’t exist without Prof. David Jones & Prof. Christine Orengo’s groups at UCL, who developed and maintain this incredible resource. Their work made this integration possible! 🏗️
@ucl.ac.uk

🧵(5/6)

None of this would have been possible without our amazing team! Huge thanks to our PDBe—their work made TED’s integration both seamless and accessible. 🙌
@pdbeurope.bsky.social

🧵(4/6)

One great example? Inosine-5'-monophosphate dehydrogenase (AF-Q9GZH3-F1), where TED highlights a CBS-domain nested inside a split Aldolase class I domain. TED annotations + the interactive PAE plot = a powerful way to explore domain organisation and relationship!

🧵(3/6)

TED (The Encyclopedia of Domains) makes it easier to explore structural domains in AlphaFold-predicted proteins. You can visualise functional units, assess domain organisation with the PAE plot, and explore evolutionary relationships through CATH classifications (@cathgene3d.bsky.social).

🧵(2/6)

AlphaFold DB now has TED integrated! This is a special one for me — my first project where I took ownership from start to finish, bringing together different teams to integrate domain annotations into #AlphaFold DB. A huge learning experience, and I’m excited to finally see it live!

🧵(1/6)

Wow, these are all so incredible! Especially the DNA lego bases! I'd love to learn to do this more. Do you have any tips to get started? I know the main bottleneck for me is I don't have GPU rendering capabilities...