Latest news with #JohnJumper
Forbes
3 hours ago
- Health
- Forbes
Alphafold 3 Extends Modeling Capacity To More Biological Targets
Doctor working on digital tablet with medical interface and digital healthcare and network concept. The people behind the original protein modeling tool Alphafold have now developed a newer version, Alphafold 3, which is changing the way that this fundamental technology works. Looking at the changes in the newest version, you find that Alphafold 3 extends to a broader spectrum of molecular structures, including ligands (ions or molecules with certain binding properties) and other ions, as well as something called 'post-translational modifications' – (here's the Wikipedia entry.) Additionally, Alphafold 3 uses a reformed 'Pairformer' architecture to process pairwise relationships (more on that later) - it has better prediction accuracy, and improved performance in making some types of predictions. (Here's more from NIH). The original Alphafold technology earned its makers, John Jumper and Demis Hassabis, a Nobel prize, and now these tools are still redefining what it means to do drug discovery. So how does Alphafold work for big drug companies? In a TED talk explaining some of this commercial success, Lauren Davis, someone with MIT ties, shows us a bit of how this works, helping companies to come up with life-saving medicines. Davis points to a 'transformative' process where new tools enable rapid development, on a more efficient framework. One aspect of this, she points out, is target identification – predicting the structure of a given target. That way, companies can sidestep some of the human and animal testing that's expensive and labor-intensive, not to mention sensitive. She compares the system to a dating app, where you narrow down potential matches before proceeding with actual testing, which would be analogous to meeting someone for a date (read: investing time and effort.) She talks about the process of scoring potential inhibitors, which she says she's excited about because she used to be on the MIT soccer team. In general, Davis paints a picture of how Alphafold actually applies in the commercial world. That's a little bit about the way is that Alphafold is contributing to the medical community. But there's another feature of this new model that I was interested in, and I ended up getting different results from different LLM models, so let's look at that a bit: If you ask ChatGPT whether Alphafold 3 is open source, you get this - at least, I did: '(Alphqafold 3 is) not open-sourced; instead, it is accessible through a cloud-based platform provided by DeepMind for non-commercial research purposes. This approach has elicited some concerns within the scientific community regarding transparency and accessibility.' As a source, the model lists this Wired article. However, when I looked at a post from Dario Amodei from November of last year, I found this, suggesting, albeit in a terse way, that Alphafold 3 has become open source: 'AI protein prediction tool, AlphaFold3, is open source.' Now, when I asked Copilot the same thing, as a result of typing a search into Bing, it gave me this: 'Not fully open source - AlphaFold 3 is not fully open source. While the source code and model weights are available for academic use under specific non-commercial restrictions, access to the model weights is limited to those with academic affiliations. This means that while researchers can use the software for non-commercial applications, they cannot freely access the training weights for commercial use.' And there were a number of sources listed. So the most likely answer, based on all of that input, is that Alphafold 3 is 'sort of' open source – that is, as Copilot said, that some of the weights and other aspects are public, but other aspects of the technology are not. ChatGPT contends, above, that the cloud-based platform that Alphafold 3 is on is 'maintained by DeepMind for non-commercial research purposes.' Anyway, in this case, we don't really have to guess: just check the GitHub for the new version. But this shows how you can get different information from different models, something we are going to have to learn to navigate. In any case, this new version of Alphafold continues the tradition of giving us new tools for drug discovery. Davis, in her talk, used the example of lisinopril as an ACE inhibitor, and explained how the Alphafold process can apply to predicting and modeling how an inhibitor will work. Practically, we have millions and millions of Americans on these drugs, and they apply to a wide scope of health conditions, so it's abundantly useful to take advantage of AI in these ways. Will it lower the cost of drugs? We'll see.
Forbes
4 days ago
- Science
- Forbes
The Prototype: This AI Model Could Make It Faster To Find New Medicines
In this week's edition of The Prototype, we look at a new AI model that could speed up drug discovery, how the Trump/Musk blowup could impact NASA, a new class of electronics and more. You can sign up to get The Prototype in your inbox here. getty The 2024 Nobel Prize in Chemistry was awarded in part to Deepmind's Demis Hassabis and John Jumper for the development of AlphaFold–an AI model that predicts the structure of proteins, the complex chemicals essential to making our bodies work. Since its inception, this model and others like it have been put to use in laboratories around the world, enabling new biological discoveries. Now a team from MIT and pharmaceutical company Recursion, with support from Cancer Grand Challenges, have developed a tool that takes these principles further–and may help researchers find new medicines more quickly. Called Boltz-2, this open-source generative AI model can not only predict the structure of proteins, it can also predict its binding affinity–that is, how well a potential drug is able to interact with that protein. This is crucial in the early stages of developing a new medicine. Until now, scientists could only find binding affinity in one of two ways: they could actually conduct an experiment to determine it, or they could use a complicated computer simulation process called FEP. In a paper published today, which has not yet been peer-reviewed, the team demonstrated that it could produce similar results to an FEP–but significantly faster. 'Boltz-2, in just 20 seconds, can match the performance of FEP, which usually takes from 6-12 hours,' said researcher Gabriele Corso. 'Pretty much changing the game.' Getty Images SpaceX has been caught in the crossfire of the ongoing feud between Donald Trump and company founder Elon Musk. The two men have been sharing barbs over the President's proposed budget bill, with Musk criticizing it for including too much spending and increasing the deficit. On Thursday afternoon, the President posted on Truth Social that '[t]he easiest way to save money in our Budget, Billions and Billions of Dollars, is to terminate Elon's Governmental Subsidies and Contracts.' If Trump were to follow through on cancelling contracts, the biggest price may well be paid by NASA. Although the space agency played a crucial role in getting the company off the ground, SpaceX doesn't need it anymore. According to Musk, the company is currently bringing in around $15.5 billion a year in revenue. Forbes estimates that about 80% of this comes from its internet business, Starlink. And while SpaceX still gets plenty of government business, it also launches dozens of commercial spacecraft every year. The reverse, however, isn't true. NASA relies heavily on SpaceX for its operations–the company's rockets launched more than half of the agency's space missions last year. And while NASA has other partners in aerospace, such as Boeing, many are years behind SpaceX in terms of development. Read the whole story here. A team of researchers at Virginia Tech invented a new kind of circuit board that is both more durable and easier to recycle than conventional electronics. It's composed of a soft plastic that's embedded with a liquid, conductive metal to carry electricity. The resulting electronics work even if they're bent out of shape and can even self-repair some damage. For recycling, they can be chemically deconstructed with a simple process that makes it easy to re-form into a new product. Japanese space startup Ispace's second attempt to land a spacecraft on the Moon failed this week. According to the company, the laser rangefinder that its spacecraft used to measure the distance to the surface experienced communications difficulties. Because it didn't know its altitude, it didn't slow down enough for a safe landing, causing it to crash. In my other newsletter, InnovationRx, Amy Feldman and I looked at the impact of Trump's proposed budget cuts on biomedical research and global health, news from the ASCO cancer meeting and a biotech company making drug products through fermentation. Solar panels provide an unexpected environmental benefit–when they're placed in drought-prone grasslands, they boost soil moisture levels and increase plant growth by 20% compared to open fields, because of both the shade they provide and water that collects on them. A new compound, called infuzide, showed antibacterial activity against strains that are resistant to antibiotics, which may provide a new weapon for doctors against infectious diseases. Amazon is reportedly testing humanoid robots to see if they can be used to deliver packages. The retail giant has already been putting similar technology to work in its warehouses. Researchers found that diatoms, a kind of algae with silica in its cell walls, could be introduced to the Moon's soil to make it capable of growing crops. If you're in midlife and wondering if you should abandon your morning coffee, think twice–at least, if you're a woman. That's because a new analysis, which followed nearly 50,000 women for over 30 years, found that those who drank coffee (the good stuff, with caffeine) were more likely to age in a healthy way, maintaining both their physical and cognitive health across a wide variety of parameters, than those who drank tea or decaf. As a middle-aged dad, two things I greatly enjoy are hard rock music and military history. Swedish metal band Sabaton scratches both of those itches at the same time by singing heavy ballads about historic battles and the people who fought them. Some of my favorite tracks of theirs include 'Night Witches' (about an all-female Soviet bomber regiment in World War II), 'The Last Stand' (about the Swiss Guards who defended Rome in battle in 1527), "Blood of Bannockburn" (about a major victory in the War of Scottish Independence) and 'To Hell And Back' (about American World War II hero Audie Murphy). They're like Schoolhouse Rock but with much better guitar solos.
Yahoo
7 days ago
- Business
- Yahoo
Nvidia, Dell announce major project to reshape AI
Nvidia, Dell announce major project to reshape AI originally appeared on TheStreet. I believe that the universe always keeps things in balance. For every positive thing, there is a negative, and vice versa. Imagine working as a teacher for a moment. The world has changed, and suddenly everyone has access to artificial intelligence. Are your students using ChatGPT to do their homework? Absolutely. Would you like to be in that teacher's shoes? I know I wouldn't. What if this AI revolution turns out to be a tragedy like the use of leaded petrol, which is suspected to have lowered the IQ of Americans born in the 1960s and 1970s? While AI advances could potentially extinguish future scientific minds, today's scientists use powerful computers to deliver scientific breakthroughs. Google's AlphaFold, a program for protein structure prediction, had already made breakthroughs in 2018 before the advent of agentic AI. In 2024, its authors Demis Hassabis and John Jumper were awarded one-half of the Nobel Prize in Chemistry, the other half went to David Baker for his work on protein design. Baker wasn't doing his research on pen and paper either; he relied on the National Energy Research Scientific Computing Center's Perlmutter supercomputer to do his work. Now, Dell is working on something for those for whom Perlmutter isn't good enough. Dell Technologies () released its earnings report for Q1 Fiscal 2026 on May 29. Here are some of the highlights: Revenue of $23.4 billion, up 5% year over year Operating income of $1.2 billion, up 21% YoY Diluted EPS of $1.37, flat YoY, 'We achieved first-quarter record servers and networking revenue of $6.3 billion, and we're experiencing unprecedented demand for our AI-optimized servers. We generated $12.1 billion in AI orders this quarter alone, surpassing the entirety of shipments in all of FY25 and leaving us with $14.4 billion in backlog," stated Jeff Clarke, vice chairman and chief operating officer of Dell. Most of that backlog consists of complex systems built using Nvidia () Blackwell Dell is leaning heavily on Nvidia, Nvidia is looking for ways to minimize losses caused by new government policies that require a license to export its H20 chip to China. As TheStreet's Samuel O'Brient reports, Nvidia could not ship an additional $2.5 billion worth of H20 products during Q1 because of the restrictions. On top of that, Nvidia expects the H20 licensing requirement to result in an $8 billion revenue hit during Q2. Nvidia's guidance is for roughly $45 billion in sales in the second quarter. On May 29, Nvidia and Dell announced Doudna, a supercomputer for NERSC, a U.S. Department of Energy user facility at Berkeley Lab. It is set to launch in 2026 and is named for Nobel laureate and CRISPR pioneer Jennifer Doudna. According to Nvidia, Doudna is expected to outperform its predecessor, Perlmutter, by more than 10x in scientific output, all while using just 2-3x the power. It will be powered by NVIDIA's next-generation Vera Rubin chips.'I'm so proud that America continues to invest in this particular area,' stated NVIDIA founder and CEO Jensen Huang. 'It is the foundation of scientific discovery for our country. It is also the foundation for economic and technology leadership.' More Nvidia: Analysts issue rare warning on Nvidia stock before key earnings Analysts double price target of new AI stock backed by Nvidia Nvidia CEO shares blunt message on China chip sales ban Unlike conventional systems, Doudna merges simulation, data, and AI into a single seamless platform, built for real-time discovery. 'We're not just building a faster computer,' stated Nick Wright, advanced technologies group lead and Doudna chief architect at NERSC. 'We're building a system that helps researchers think bigger and discover sooner.' Doudna includes support for scalable quantum algorithm development and the co-design of future integrated quantum high-performance computing systems. Research teams, working on climate models and particle physics, are already porting full workflows to Doudna. Nvidia seems to be finding ways to recoup the revenue losses created by the new regulations, as Huang recently hinted at the possibility of greater partnership with Tesla and Dell announce major project to reshape AI first appeared on TheStreet on Jun 3, 2025 This story was originally reported by TheStreet on Jun 3, 2025, where it first appeared. Error in retrieving data Sign in to access your portfolio Error in retrieving data Error in retrieving data Error in retrieving data Error in retrieving data
Miami Herald
7 days ago
- Business
- Miami Herald
Nvidia, Dell announce major project to reshape AI
I believe that the universe always keeps things in balance. For every positive thing, there is a negative, and vice versa. Imagine working as a teacher for a moment. The world has changed, and suddenly everyone has access to artificial intelligence. Are your students using ChatGPT to do their homework? Absolutely. Would you like to be in that teacher's shoes? I know I wouldn't. What if this AI revolution turns out to be a tragedy like the use of leaded petrol, which is suspected to have lowered the IQ of Americans born in the 1960s and 1970s? While AI advances could potentially extinguish future scientific minds, today's scientists use powerful computers to deliver scientific breakthroughs. Google's AlphaFold, a program for protein structure prediction, had already made breakthroughs in 2018 before the advent of agentic AI. In 2024, its authors Demis Hassabis and John Jumper were awarded one-half of the Nobel Prize in Chemistry, the other half went to David Baker for his work on protein design. Baker wasn't doing his research on pen and paper either; he relied on the National Energy Research Scientific Computing Center's Perlmutter supercomputer to do his work. Now, Dell is working on something for those for whom Perlmutter isn't good Technologies (DELL) released its earnings report for Q1 Fiscal 2026 on May 29. Here are some of the highlights: Revenue of $23.4 billion, up 5% year over yearOperating income of $1.2 billion, up 21% YoYDiluted EPS of $1.37, flat YoY, "We achieved first-quarter record servers and networking revenue of $6.3 billion, and we're experiencing unprecedented demand for our AI-optimized servers. We generated $12.1 billion in AI orders this quarter alone, surpassing the entirety of shipments in all of FY25 and leaving us with $14.4 billion in backlog," stated Jeff Clarke, vice chairman and chief operating officer of Dell. Most of that backlog consists of complex systems built using Nvidia (NVDA) Blackwell chips. Related: Dell execs sound alarm with consumer comments While Dell is leaning heavily on Nvidia, Nvidia is looking for ways to minimize losses caused by new government policies that require a license to export its H20 chip to China. As TheStreet's Samuel O'Brient reports, Nvidia could not ship an additional $2.5 billion worth of H20 products during Q1 because of the restrictions. On top of that, Nvidia expects the H20 licensing requirement to result in an $8 billion revenue hit during Q2. Nvidia's guidance is for roughly $45 billion in sales in the second quarter. On May 29, Nvidia and Dell announced Doudna, a supercomputer for NERSC, a U.S. Department of Energy user facility at Berkeley Lab. It is set to launch in 2026 and is named for Nobel laureate and CRISPR pioneer Jennifer Doudna. According to Nvidia, Doudna is expected to outperform its predecessor, Perlmutter, by more than 10x in scientific output, all while using just 2-3x the power. It will be powered by NVIDIA's next-generation Vera Rubin chips. Related: Popular cloud storage service might be oversharing your data "I'm so proud that America continues to invest in this particular area," stated NVIDIA founder and CEO Jensen Huang. "It is the foundation of scientific discovery for our country. It is also the foundation for economic and technology leadership." More Nvidia: Analysts issue rare warning on Nvidia stock before key earningsAnalysts double price target of new AI stock backed by NvidiaNvidia CEO shares blunt message on China chip sales ban Unlike conventional systems, Doudna merges simulation, data, and AI into a single seamless platform, built for real-time discovery. "We're not just building a faster computer," stated Nick Wright, advanced technologies group lead and Doudna chief architect at NERSC. "We're building a system that helps researchers think bigger and discover sooner." Doudna includes support for scalable quantum algorithm development and the co-design of future integrated quantum high-performance computing systems. Research teams, working on climate models and particle physics, are already porting full workflows to Doudna. Nvidia seems to be finding ways to recoup the revenue losses created by the new regulations, as Huang recently hinted at the possibility of greater partnership with Tesla and xAI. Related: Veteran fund manager who predicted April rally updates S&P 500 forecast The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc.
The Guardian
28-03-2025
- Science
- The Guardian
AI may help us cure countless diseases – and usher in a new golden age of medicine
AlphaFold might be the most exciting scientific innovation of this century. From Google DeepMind, and first reported in 2020, it uses artificial intelligence to figure out a protein's 3D structure. The technology has already been used to solve fundamental questions in biology, awarded the Nobel prize (in chemistry – to Demis Hassabis and John Jumper) and revolutionised drug discovery. Like most AI, it's only getting better – and just getting started. A protein's structure gives us clues about its function, and helps us design new drugs. AlphaFold, which was trained on a huge database of experimentally solved structures called the Protein Data Bank, predicts a protein's structure based on its amino acid sequence. In the past, the first step would be to produce a vast amount of protein – using litres of a bacterium, or a virus. You'd pray for the protein to assemble into a crystal lattice (notoriously difficult), and then fire high-energy X-rays at it. This is called X-ray crystallography, and it could take years. Now, AlphaFold can do it in minutes (and a hell of a lot more cheaply, too). When AlphaFold competed at a protein structure-solving competition in 2020, it was so good that some accused the AlphaFold team of cheating. At its first appearance, AlphaFold became the state of the art. Now, there are approximately 250,000,000 protein structures in the AlphaFold database, which has been used by almost 2 million people from 190 countries – many more people than can do X-ray crystallography! I did my PhD on cancer biology. I would have loved to solve the structure of the protein I worked on. Maybe I could have even used it to make a new drug. Now, I can go to the AlphaFold server and produce a structure in five minutes that would have consumed my whole PhD. Dr Pauline Lascaux, a molecular and structural biologist at the University of Oxford, said that AlphaFold was central to her latest study, which discovered a new way that cells repair DNA, and that more than 90% of the studies she reviews are citing it. So what does drug discovery with AlphaFold look like? In this recent Science study, researchers used AlphaFold to predict the structure of the serotonin receptor, which controls mood. Through in silico testing (on computers) they tested which of 1.6bn (!) molecules could bind the AlphaFold structure. What they found was a series of molecules that bound much more tightly than drugs generated via the conventional – experimental – approach, which could be new drugs for mood disorders. AlphaFold has only been around since 2020, but its impact has been meteoric. Here are the top three discoveries enabled by AlphaFold so far: Solving a decades-old problem: the structure of the nuclear pore complex, one of the biggest structures in the cell. This complex is the guardian of entry to the nucleus, which holds the cell's DNA. It's implicated in cancer, ageing and neurodegeneration – and now we know what it looks like at the atomic level. Finding a new liver cancer drug. In a lab (not in patients), the drug, which targets the cancer protein CDK20, prevented liver cancer growth. Helping to design a 'molecular syringe', which delivers a therapeutic protein payload into human cells. There are companies built on AlphaFold, too. If AlphaFold is solving the lock, then AlphaProteo provides the key. AlphaProteo uses AlphaFold's structures to design molecules that can bind to and modulate other proteins. This has been used to generate molecules that have never been made before – to target Covid-19, cancer and autoimmunity. Also from DeepMind, AlphaMissense tackles the problem of missense mutations – minor changes to genes, with uncertain functional impact. Despite their prevalence, we only know whether about 2% of these changes are pathogenic. AlphaMissense models the structure of the mutations using AlphaFold: if the protein structure changes, it's probably pathogenic. This could transform the diagnosis and treatment of rare genetic diseases. We don't know yet whether drugs designed using AlphaFold will pass successfully through clinical trials, and none have been tested in humans yet – only time will tell. In the future, AlphaFold could enable new medicines to be discovered by individuals, could find drugs for undruggable targets, and could unlock the secrets of molecular life. (Just the other day, AlphaFold helped to solve the structure of the sperm-egg bridge that forms during fertilisation.) If the first generation of drug discovery was the nature generation, which gave us aspirin (from willow tree bark), and the second was the biotech generation, which gave us Ozempic, then we've now moved to the third generation: the AI generation. Samuel Hume is a fellow at The Foulkes Foundation and pursuing PhD in the University of Oxford's department of oncology
