Much of the recent hype about AI has focused on the fascinating digital content generated from simple prompts, alongside concerns about its ability to decimate the workforce and make malicious propaganda much more convincing. (Fun!) However, some of the most promising – and potentially much less worrying – work in AI lies in medicine. A new update to Google’s AlphaFold software could lead to new advances in disease research and treatment.
AlphaFold software, from Google DeepMind and Isomorphic Labs (also owned by Alphabet), has already been shown to predict how proteins fold with shocking accuracy. It lists 200 million known proteins, and Google says millions of researchers have used previous versions to make discoveries in areas such as malaria vaccines, cancer treatment and enzyme design.
Knowing the shape and structure of a protein determines how it interacts with the human body, allowing scientists to create new drugs or improve existing ones. But the new version, AlphaFold 3, can model other crucial molecules, including DNA. It can also map interactions between drugs and diseases, which could open exciting new doors for researchers. And Google claims to do it with 50% greater accuracy than existing models.
“AlphaFold 3 takes us beyond proteins to a broad spectrum of biomolecules,” Google’s DeepMind research team wrote in a blog post. “This breakthrough could pave the way for more transformative science, from developing biorenewable materials and more resilient crops to accelerating drug design and genomics research.” »
“How do proteins respond to DNA damage?” how do they find it, repair it? John Jumper, Google DeepMind project leader, said Wired. “We can start to answer these questions.”
Before AI, scientists could only study protein structures using electron microscopes and elaborate methods like X-ray crystallography. Machine learning streamlines much of this process by using patterns recognized during its training (often imperceptible to humans and our standard instruments) to predict the shapes of proteins based on their amino acids.
Google says part of AlphaFold 3’s advances come from applying diffusion models to its molecular predictions. Diffusion models are central elements of AI image generators like Midjourney, Google’s Gemini, and OpenAI’s DALL-E 3. Integrating these algorithms into AlphaFold “refines the molecular structures generated by the software,” as Wired explain. In other words, it takes training that seems fuzzy or vague and makes very educated guesses based on patterns from its training data to clear it up.
“This is a big step forward for us,” said Demis Hassabis, CEO of Google DeepMind. Wired. “This is exactly what you need for drug discovery: you need to see how a small molecule is going to bind to a drug, how strongly, and what else it might bind to.”
AlphaFold 3 uses a color-coded scale to indicate its level of confidence in its prediction, allowing researchers to err on the side of caution with results less likely to be accurate. Blue means high confidence; red means it is less secure.
Google is making AlphaFold 3 free for researchers to use for non-commercial research purposes. However, unlike previous releases, the company is not making the project open source. A prominent researcher who creates similar software, Professor David Baker of the University of Washington, expressed disappointment with Wired that Google has chosen this path. However, he was also attracted by the capabilities of the software. “The structure prediction performance of AlphaFold 3 is very impressive,” he said.
As for what’s next, Google says that “Isomorphic Labs is already collaborating with pharmaceutical companies to apply it to real-world drug design challenges and ultimately develop new treatments that change patients’ lives.” »
News Source : www.engadget.com
Gn tech
