OpenFold3-preview (OF3p) is out: a sneak peek of our AF3-based structure prediction model. Our aim for OF3 is full AF3-parity for every modality. We now believe we have a clear path towards this goal and are releasing OF3p to enable building in the OF3 ecosystem. More👇
You can get MSAs directly from AlphaFold DB now (alphafold.ebi.ac.uk). I also missed the FoldSeek implementation, which lets you search for structurally similar proteins direktly.
This saves some clicking around. Neat!
I get that the news cycle is packed right now, but I just heard from a colleague at the Smithsonian that this is fully a GIANT SQUID BEING EATEN BY A SPERM WHALE and it’s possibly the first ever confirmed video according to a friend at NOAA
10 YEAR OLD ME IS LOSING HER MIND (a thread 🧵)
He's making a list, he's checkin' it twice
He's gonna find out who's naughty or nice
RW Official Sleuth Whitelist is out!
If you aren't on it, you are trash.
retractionwatch.com/2018/06/17/m...
Fantastic new paper by @reeserichardson.bsky.social et al.
An enormous amount of work showing the extent of coordinated scientific fraud and involvement of some editors.
The number of fraudulent publications grows at a rate far outpacing that of legitimate science.
www.pnas.org/doi/10.1073/...
PSA = Google Colab Pro free for one year for academic use (US only):
blog.google/outreach-ini...
Thanks! But I think you got the wrong person.
Zero-shot antibody design in a 24-well plate
Figure 1
Figure 2
Figure 3
Zero-shot antibody design in a 24-well plate [new]
Achieves high hit rates for de novo antibody/nanobody and miniprotein design, validated experimentally against diverse targets without prior binders.
Comparison of inverse folding models trained on PDB, AFDB, and debiased AFDB. Debiasing the AFDB leads to partial recovery of performance in every architecture shown here
Predicted structures from the AFDB are poor substitutes for experimental structures when training inverse folding NNs. Here, authors show that this is because they are "too perfect", and passing them through a custom-trained AFDB-to-PDB NN helps fix this
threadreaderapp.com/thread/19384...
I listened to the @rorinstitute.bsky.social talk about Funding by Algorithm bsky.app/profile/rori...
I think it's arrogant and dangerous to suppose that AI has a role in assessing grants. @astrokatie.com says why, very clearly.
NEW: A federal judge has ruled that the Trump administration’s termination of hundreds of NIH grants was “void and illegal.”
He ordered funding for some of the grants, including many profiled by ProPublica in recent months, to be reinstated.
By @anniewaldman.bsky.social
“The search is on for 25 researchers from around the world in Lund University’s largest ever international recruitment drive. The main focus will be on AI research and other areas in which the University has a strategic focus.”
www.lunduniversity.lu.se/article/lund...
Thanks Ian for your leadership and great meeting you today at NIH!
PSA: If you're trying to run the Boltz ColabFold notebook and finding that it's crashing with this error, it's because you need to switch to an A100 GPU; the notebook no longer works with T4/L4 GPUs
A case study on the challenges of evaluating AI predictions in biology and the implications for published results.
1/2
rachel.fast.ai/posts/2025-0...
It’s not greed that drives the world, but envy. Charlie Munger.
Our work on deriving protein-protein (functional) associations across different human tissues from proteomics data is now published. There is quite a lot in this paper, so I will just touch on a few highlights:
www.nature.com/articles/s41...
I am shocked by the news coming from the NIH. I was a postdoc there and know many of the PIs affected at NINDS. Don’t be naive, this is a deliberate attack meant to cripple the system. No scientist is safe, regardless of political stance or whether you have remained neutral
Butterflies on a blue sky
THREAD
The numbers are in. @bsky.app research sharing volumes vs X Formerly Twitter
In March 2024, on most days, Bluesky hosts more posts linked to research published in 2025 than X.
By quite a lot.
Release the Kraken...
#AcademicSky #HigherEd #Altmetrics
1/11
Roses are red
Violets are blue
FREEDOM OF EXPRESSION FOR HETEROCHROMATIN
#standupforscience
Journals: We will not publish your paper because if insufficient novelty
Also journals: Congrats, your paper has been accepted. Please remove any instance of the words “new” and “novel”
DeepMind patents structure prediction. www.freepatentsonline.com/y2025/006970...
🚀 New in AlphaFold DB: Structural domains at a glance! 🔍✨
AFDB now integrates TED (The Encyclopedia of Domains) to classify functional protein domains.
🔗 More
ebi.ac.uk/about/news/updates-from-data-resources/alphafold-database-ted
🔗 Explore
alphafold.ebi.ac.uk
#AlphaFold #TED #ProteinStructures
We applaud @predatoryjournals.org for their list of 2,775 'predatory' journals (as at 01 Mar 2025).
We can only imagine the work that is involved in maintaining such a list. We have a few (we hope constructive) comments that we hope influence their development decisions going forward.
Nice resource! #AlphaFold #PPI #ProteinInteractions
Predictomes, a classifier-curated database of AlphaFold-modeled protein-protein interactions: Molecular Cell www.cell.com/molecular-ce...
