Google DeepMind's revolutionary protein-folding AI, AlphaFold3, has been a game-changer in the field of structural biology. However, its availability to researchers and pharmaceutical companies has sparked debate over accessibility and commercialization.
A new open-source artificial intelligence model, OpenFold3, aims to match AlphaFold3's performance. Developed by a non-profit collaboration of academic and private research groups, OpenFold3 uses proteins' amino acid sequences to map their 3D structures and model how they interact with other molecules.
The system was trained on over 300,000 molecular structures and a synthetic database of more than 40 million structures. Developing it has cost around $17 million so far. Unlike AlphaFold3, which is available for restricted academic use, OpenFold3 is open to any researcher or pharmaceutical company.
OpenFold3's creators have released the system as a "sneak preview" to give researchers and industry professionals a taste of its capabilities. The consortium team plans to refine the model further before releasing it fully, with the goal of reaching parity with AlphaFold3.
While OpenFold3 still lacks some of AlphaFold3's features, its open-source nature is seen as a significant step forward in democratizing AI structural-biology tools. Researchers are already excited about testing and integrating OpenFold3 into their workflows.
The development of open-source models like OpenFold3 comes at a time when Google DeepMind initially shared the underlying code for AlphaFold3 without much fanfare. Critics argued that this move limited access to cutting-edge technology. However, the company later made the AlphaFold3 code and model weights available to academics, although they remain unavailable for commercial use.
The emergence of OpenFold3 signals a new era in collaboration and innovation among researchers and industry players. As researchers continue to test and refine the system, it will be interesting to see how it compares to existing models and what potential applications it might unlock.
A new open-source artificial intelligence model, OpenFold3, aims to match AlphaFold3's performance. Developed by a non-profit collaboration of academic and private research groups, OpenFold3 uses proteins' amino acid sequences to map their 3D structures and model how they interact with other molecules.
The system was trained on over 300,000 molecular structures and a synthetic database of more than 40 million structures. Developing it has cost around $17 million so far. Unlike AlphaFold3, which is available for restricted academic use, OpenFold3 is open to any researcher or pharmaceutical company.
OpenFold3's creators have released the system as a "sneak preview" to give researchers and industry professionals a taste of its capabilities. The consortium team plans to refine the model further before releasing it fully, with the goal of reaching parity with AlphaFold3.
While OpenFold3 still lacks some of AlphaFold3's features, its open-source nature is seen as a significant step forward in democratizing AI structural-biology tools. Researchers are already excited about testing and integrating OpenFold3 into their workflows.
The development of open-source models like OpenFold3 comes at a time when Google DeepMind initially shared the underlying code for AlphaFold3 without much fanfare. Critics argued that this move limited access to cutting-edge technology. However, the company later made the AlphaFold3 code and model weights available to academics, although they remain unavailable for commercial use.
The emergence of OpenFold3 signals a new era in collaboration and innovation among researchers and industry players. As researchers continue to test and refine the system, it will be interesting to see how it compares to existing models and what potential applications it might unlock.
OPENFOLD3 IS GOING TO BRING DOWN PRICES FOR STRUCTURAL BIOLOGY SIMULATIONS AND GIVE MORE PEOPLE ACCESS TO THIS KIND OF AI POWER. IT'S ALSO COOL TO SEE NON-PROFIT GROUPS TAKING THE LEAD ON DEVELOPING OPEN-SOURCE TECH THAT COULD CHANGE THE GAME FOR PHARMA AND RESEARCHERS ALIKE! $17 MILLION IS A LOT, BUT I GUESS YOU GOTTA SPEND MONEY TO MAKE GOOD THINGS HAPPEN. NOW WE JUST WAIT AND SEE HOW IT COMPARES TO ALPHA FOLD3 AND WHAT KIND OF PROGRESS WE CAN EXPECT FROM THIS NEW PLAYER! 
, can't they just make more open-source stuff? like, google deepmind is all about innovation but then makes some stuff exclusive... OpenFold3 seems legit tho 
, training on 300k+ structures and 40m+ synthetic structures is wild! i'm curious to see how it compares to AlphaFold3 in the future 
! It's like a game-changer for researchers and pharmaceutical companies who want to work with protein-folding AI without having to shell out millions of dollars 
. And the fact that it's open-sourced is, like, super cool 
, because now more people can access this tech and help advance the field.
, but I am happy they finally shared their code (albeit with some restrictions 
). And now we have OpenFold3 to compete with it! 
The fact that these non-profit teams are working together and sharing resources is, like, so inspiring 
. Can't wait to see what kinda cool stuff researchers come up with using this tech 
I mean think about it we're living in an age where AI is advancing so fast and making breakthroughs like protein folding way back into biology its mind-blowing 
and we stopped at this quirky little cafe that had the best vegan burgers I've ever tasted 
. The atmosphere was so chill, like they wanted to make sure everyone felt welcome and relaxed. Anyway, I started thinking about how protein-folding AI is like trying to find a specific song on a vast music streaming platform... you gotta navigate through all this info to stumble upon the right one 
. And then I realized that's kinda similar to how researchers are gonna dig into OpenFold3 and figure out its potential applications 
- it's like a breath of fresh air for researchers who can't afford (or don't want to pay) AlphaFold3's hefty prices 
. But, at the same time, I'm also worried that it might create some disruption in the field and we're not entirely sure what kind of "parity" these open-source models will reach with AlphaFold3 
. Still, the fact that researchers can now use and build upon this tech without having to jump through hoops is a big deal 
they're trying 2 make ai structural biology tools more accessible 4 EVERYONE!!! i mean who wouldn't wanna predict protein structures 4 a measly $100k 
it's crazy to think about how much google deepmind spent on alphafold3 & now they're releasing the tech 2 academics anyway 
i wish more companies would follow suit & make science accessible 4 everyone! 
. OpenFold3 is paving the way for a more inclusive research community, where collaboration and innovation know no borders. Can't wait to see what kind of breakthroughs come out of this 
think its awesome that some genius group created an AI model that can already beat AlphaFold3 in some cases, and now they're making it open-source 
. I mean, $17 million is no chump change, and you can't just develop something that good without some serious resources 
. It'll be interesting to see how it pans out...
. And let's not forget the 300k+ molecular structures it was trained on 
. I mean, $17 million might seem like a lot, but think about all the potential breakthroughs that could come from this kind of collaboration... it's like a game-changer for science! 
. We're seeing these massive AI advancements being pushed out to researchers & industries, but we gotta remember that access isn't just about the tech itself, it's also about who gets to control it 
they've been working hard with like 300k+ molecular structures & 40m+ synthetic ones which is insane! it's all about collaboration & innovation now, let's keep pushing boundaries