NEW from me @STAT:
We've learned that DeepMind is going to fund CASP, the protein structure competition that brought Google DeepMind's #AlphaFold to prominence (and a Nobel Prize), as its NIH funding runs out.
What is the key idea of #AlphaFold? There isn't one. #ismbeccb2025
After #AlphaFold 1, the team tried to work without evolutionary information, to learn what they could do from physics. Accuracy was terrible (John Jumper’s words), but they learned a lot by changing the approach. Use the plausible but false structures to create a space of possibilities to explore iteratively. “Kind of surprisingly, it started to work.” #ismbeccb2025
First keynote of #ismbeccb2025: John Jumper, first but not last surely with #MachineLearning #AI at the center.
We all know about #AlphaFold, John will tell us about the challenges along the way. Good start! “Evolution is important”
CASP, the protein structure competition that rocketed AlphaFold to fame and a Nobel, is losing its NIH funding.
David Baker on how CASP impacted his career (and Nobel Prize), and what's the latest on CASP's funding status:
Interesting post on the important limitations of #AlphaFold in helping us to truly understand protein structure and function.
Sorry, the post is on substack.
https://clauswilke.substack.com/p/no-alphafold-has-not-completely-solved
Oh man. Ich weiß auch nicht, warum die Betreuung von Promovierenden so viel Spaß macht.
Aber das gemeinsame Brainstorming und Probleme lösen, das fetzt einfach total.
Hat alles voll die Sogwirkung. Und mit jedem Paper und weiteren Tool tut sich direkt ein neues Rabbit Hole auf. 
Aktuell betreue ich eine Doktorandin, die Proteinsimulation macht.
Charité study analyzes 400 million years of enzyme evolution!
With the aid of #AlphaFold AI, a research team has decoded the laws according to which #enzymes - which catalyze chemical reactions in all living things - evolve. The results, which have now been published in the scientific journal #Nature, are relevant to the development of new antibiotics, for example.
Charité-Studie analysiert 400 Millionen Jahre Enzym-Evolution!
Mithilfe der #AlphaFold-KI hat ein Forschungsteam die Gesetzmäßigkeiten entschlüsselt, nach denen sich #Enzyme – die chemische Reaktionen in allen Lebewesen beschleunigen – weiterentwickeln. Die jetzt im Fachmagazin #Nature veröffentlichten Ergebnisse sind beispielsweise für die Entwicklung neuer Antibiotika relevant.
Google's Isomorphic Labs: Where "move fast and break things" meets "please don't break the humans" 
DeepMind's spinoff just raised $600M and is prepping human trials for AI-designed drugs. Their goal? Click a button, cure a disease. Bold move from folding proteins to potentially saving lives.
"In my view this day will not come soon,
but when it does come the X-ray crystallographers can go out of business,
perhaps with a certain sense of relief, and it will also be possible to discuss the structures of many important proteins which cannot be crystallized and therefore lie outside the crystallographer’s purview."
One wonders whether #Google's #Isomorphic labs were named after the mathematical modelling, or the isomorphous replacement technique pioneered here?
Dr Ron Vale
The Impact of AlphaFold on Cell Biology
Cellular Interfaces Seminar
Friday 20 June, 11.15am-12.15pm, GLT2, Medical School Building
Note the change of venue!
Tomorrow we are hosting a talk from Ron Vale:
The Impact of AlphaFold on Cell Biology
Dr Ron Vale, Emeritus Professor of Cellular and Molecular Pharmacology, University of California
Cellular Interfaces Seminar
Friday 20 June, 11.15am-12.15pm, A151/MBU, Medical School Building
[Перевод] Потеря работы из-за ИИ: насколько реальна угроза? Мнение Демиса Хассабиса
Имя Демиса Хассабиса — синоним амбициозных проектов в области ИИ. От разработки легендарной игры Theme Park до создания AlphaFold, решившей «нобелевскую» задачу биологии, и теперь — к AGI , искусственному интеллекту уровня человека. Глава DeepMind утверждает: шанс увидеть AGI в ближайшее десятилетие — 50/50. Этот прогноз одновременно восхищает и пугает. Ведь за ним маячит призрак массовой потери работы из‑за ИИ. Насколько реален апокалипсис профессий? И есть ли свет в конце тоннеля общего искусственного интеллекта? Погружаемся в мысли одного из главных архитекторов нашего будущего.
Improved 3D prediction from #alphafold (2/3) . Result from #casp15 
Deep-learning-based single-dome and multidomain protein structure prediction with D-I-TASSER. Nature Biotechnology https://zhanggroup.org/D-I-TASSER/help.html#casp15
New review out in Drug Discovery Today:
https://doi.org/10.1016/j.drudis.2025.104382
We take a critical look at how AI — especially AlphaFold(-Multimer/3) — has changed protein complex prediction.
End-to-end models work well for rigid complexes. Flexibility, disorder, and transient interfaces remain unsolved.
AI is shiny. Biology is complicated. We map what’s working, what isn’t, and what’s next.
#uniprot just hides the controls.. so you can't even try to full screen. but you can fortunately link out the #ebi #alphafold site where (ctd..)
The new #AlphaEvolve from #DeepMind promises a general-purpose #AI for science - from solving math puzzles to improving chip designs, maybe as revolutionary as #AlphaFold. But until an open-source version is vetted by researchers, skepticism remains the safest policy.
#Google #ScienceAI #ResearchSky
DeepMind unveils ‘spectacular’...
«AlphaFold, un ejemplo de IA que fue muy útil» https://proxy.jesusysustics.com/2025/05/13684/
Vídeo de Veritasium donde nos explican el proceso que llevó a desarrollar AlphaFold, una IA de Alphabet (Google) verdaderamente útil para la ciencia al ayudar a determinar la estructura de todas las proteínas.
Es un soplo de aire fresco recordar que los avances en ciencia también sirven para esta clase de cosas, aunque al mismo tiempo es fácil fijarse en que de esto hace cinco años y el boom de los LLM probablemente ha quitado el foco en los logros científicos para centrarse en logros comerciales.
#Alphabet #AlphaFold #Ciencia #curiosidad #InteligenciaArtificial #vídeo #YouTube
️
️ 
