🤖
07.03.2026 00:39
👍 0
🔁 0
💬 0
📌 0
🤖
Structural biologists be out here like "Maybe you should do critical assessment? No? Okay, good talk".
Context: Alphafold, critical assessment of protein structure prediction (CASP)
Out today: We discovered new viral proteins that target immune signaling molecules, solely based on their AlphaFold-predicted shapes
www.science.org/doi/10.1126/...
Congrats Nitzan Tal and coauthors! Thank you Kranzusch lab for the fun collaboration!
Linking below previous thread on our findings
If you have a protein where AlphaFold or other methods struggle (e.g., viral proteins, antibodies, or low-homology targets), give ProteinTTT a try. ProteinTTT adapts ESMFold to one protein at a time before predicting structure, boosting downstream accuracy for most proteins
#gComm
New deep learning framework RF2-PPI integrates genomic MSAs and AlphaFold structures to predict human protein interactions at scale 🧬🤖
Details: https://www.maxapress.com/article/doi/10.48130/gcomm-0026-0001
#deeplearning #education #learning #structures
In #G3journal, Peter Scharff-Poulsen and Morten Kielland-Brandt used #Alphafold to build a model of constitutively signaling and hyper-responsive mutations in #yeast SSY1 amino acid sensor, predicting confirmation dynamics responsible for signal generation.https://buff.ly/qnrdWjW
We are using the AlphaFold pooling idea applied now to yeast. link.springer.com/article/10.1.... We further optimised the process by reducing the number of recyles without a strong impact on performance. We have made predictions covering over 4 million yeast protein pairs, about 25-30% of our goal.
Congratulations, @pedrobeltrao.bsky.social. This is outstanding work.
At 12.5M pair predictions using AF3 pooling, how are you controlling for false positives driven by steric complementarity artifacts in the diffusion module versus genuine PPI signals, especially for low-MSA yeast ORFs?
We have started a project trying to predic the interactions/structures of all yeast protein pairs using an AlphaFold pooling approach. We are making the current dataset open and we welcome collaborations.
www.evocellnet.com/2026/03/mapp...
Two-panel calibration plot (two benchmark dimer datasets) comparing predicted interchain contact-probability bins (x-axis) with the observed fraction of native interfacial contacts (y-axis). Points follow the diagonal, indicating close agreement between predicted probabilities and true interface-contact fractions.
My first manuscript in MPI colours! With @tothpetroczylab.bsky.social, we show that AlphaFold PAE-derived contact probabilities are well calibrated to the fraction of true interface contacts across experimentally determined protein dimers.
www.biorxiv.org/content/10.6...
Pinc: a simple probabilistic AlphaFold interaction score
Figure 1
Figure 2
Figure 3
Pinc: a simple probabilistic AlphaFold interaction score [new]
...converts AlphaFold's predicted aligned errors into calibrated contact probabilities for screening protein interactions & prioritizing interface residues.
#IQFPaper Guiding AlphaFold predictions with experimental knowledge to inform dynamics and interactions with VAIRO. In Protein Sci
@ibmb-csic.bsky.social #universitatdebarcelona @uni-graz.at @cabd-upo-csic.bsky.social @pablodeolavide.upo.es #universitaetKonstanz #BioTechMedGraz #ICREA
bit.ly/4aLS8wP
Biggest contribution of AI to science so far: AlphaFold, which has solved problem of predicting protein 3d structure from 1d sequence.
But there are other problems of protein folding: understanding mechanisms & pathways, misfolding, & role of disordered proteins: my blogpost
softmachines.org?p=3258
The latest Retrospectiva is out:
I talk about Pi, baby carriers, Alphafold, and barbarous distillation attacks
duarteocarmo.com/blog/retros...
The problem of predicting protein structure from sequence has been definitively solved by the AI programme AlphaFold, winning a well-deserved Nobel prize for its developers.
But structure prediction is just one of at least four different problems of protein folding...
Combining information from these structures with functional information from #uniprot and co-evolutionary information via #alphafold we found new FANCD2 interactors that utilize the same site
Adversarial Sequence Mutations in AlphaFold andESMFold Reveal Nonphysical StructuralInvariance, Confidence Failures, and Concerns forProtein Design www.biorxiv.org/content/10.64898/2026.02.25.708002v1 #cryoEM
📢 Hiring a Postdoc in Computational Phage Biology (Kraków, Poland)
We study the evolution & structural modularity of prophage-encoded glycan-degrading enzymes in Klebsiella pneumoniae — combining genomics and AlphaFold-based analyses.
Details in attached PDF 👇
“The Trump administration has repeatedly condemned AI safeguards as “woke AI.”
fortune.com/2026/02/25/d...
‘An AlphaFold 4’ – scientists marvel at DeepMind drug spin-off’s exclusive new AI
www.nature.com/articles/d41...
AlphaFold modeling of polyubiquitin complexes and covalently linked proteins
AlphaFold modeling of polyubiquitin complexes and covalently linked proteins
http://dlvr.it/TR90K4
Balázs Fábián @mpibp.bsky/social & colleagues
@cp-cellrepphyssci.bsky.social
#bps2026
BioStruct-Africa’s scalable framework for AlphaFold-enabled research training and sustainable workforce development in Africa
https://www.europesays.com/africa/103999/
Our AlphaFold pilot workshop, which took place in October 2024 in Douala, Cameroon, trained 20 scientists in AlphaFold…
AlphaFold is the foremost AI for protein geometry. Deepmind used it to compress the equivalent of 200 years of research into weeks.
AlphaFold is a generative AI model. It speaks the language of protein geometry instead of english.
It's built on the same fundamental tech as LLMs & image generation.
Proteome-wide AlphaFold pool party N&V by @ksdrew.bsky.social link.springer.com/article/10.1... (I almost added a party GIF, almost :)
Nature: Drug-discovery AI is akin to ‘AlphaFold 4’
Isomorphic Labs, a biopharmaceutical spin-off of Google DeepMind. Unveiled a new powerful AI tool for predicting how proteins interact with drugs. The tool called IsoDDE, can outperform other AI systems www.nature.com/articles/d41...
New in NCI DATA : Structural Biology AI Reference Collection 🧬
A curated, versioned set of reference databases supporting protein structure prediction at scale.
Built for tools like AlphaFold 3, AlphaFold 2 and RoseTTAFold, using updated data from UniProt and the Protein Data Bank
