Artificial intelligence may be on the verge of greatly enhancing the way humans find new drugs and treat diseases.
Professor Daniel Rigden, from the University of Liverpool's School of Chemistry, Cell and Systems Biology, says new software released this month by Google's AI arm DeepMind has the potential to change the way new drugs are designed, allowing them to more effectively target diseases. He said it was possible.
Rigden said AlphaFold3 allows scientists to open a “window” into biological processes by predicting how molecules interact. “This program could be particularly important in the development of new drugs because it can accurately predict how a drug target (usually a protein) will bind to a drug (usually a small organic molecule),” he said. .”It has said.
How does it work?
This predicts how different proteins interact, and how they (uniquely) interact with DNA and RNA, Rigden explains.
AlphaFold3 is superior to other methods, he says, and the new version performs better and can perform a wider range of predictive tasks than its predecessor, AlphaFold2.
A Google spokesperson told Yahoo News: “AlphaFold 3 is a revolutionary AI model that can predict the structure and interactions of every molecule in life. “It helps us visualize things, so we can build better systems.” Understanding important processes in the human body, other species, and even plants.
“AlphaFold 3 is a quantum leap in accuracy over Google DeepMind's previous AlphaFold 2 model, with significantly improved accuracy for key biomolecular classes.”
What can it accomplish?
Ridgin said Alphafold3's importance stems from its ability to accurately predict the structure of molecules and their interactions.
Rigden says Alphafold3 can also be useful in “literally any field of biological research.”
Google says the software could help with everything from finding effective new treatments and tackling disease to tackling food security challenges by tackling viruses that affect plants. Suggests. It could also help us understand how our DNA is read, copied, and repaired so that we continue to thrive as a species.
Google Deepmind's Isomorphic Labs is already using AlphaFold3 to discover new drugs.
A Google spokesperson said: “Combined with other AI tools developed at Isomorphic Labs, we can accelerate (by reducing the time required for drug discovery) and (better understand and characterize disease targets). “We believe that this has the potential to improve and ultimately bring about change.” drug discovery (by enabling the pursuit of entirely new targets that benefit different patient populations);
“This research has the potential to accelerate our understanding of the molecular machinery that powers the human body and unlock a world of possibilities in areas such as plant immunity, drug design, understanding biorenewable materials, and exciting genomics research.” is hidden.”
Why the difference?
Determining protein structure experimentally (which remains the “gold standard” for what AlphaFold3 does) can cost up to $100,000.
Therefore, being able to predict the structure of your software has the potential to save you a huge amount of time and money.
AlphaFold2's database of 200 million protein structure predictions is already widely used.
A Google spokesperson said: “The platform's cutting-edge predictions cover nearly every type of molecule in the Protein Databank, which stores all known biomolecular structures solved by painstaking experiments. “There is,” he said.
“In conjunction with AlphaFold 3, Google DeepMind also launched AlphaFold Server, a web-based tool that allows scientists to generate predictions of these cell interactions with just a few clicks of a button. “You can advance this important scientific research for free.'' ”
Dr Nicole Wheeler, Birmingham Fellow at the Institute of Microbial and Infectious Diseases at the University of Birmingham, said: Model complexity. This is an important advance as the US government begins imposing reporting requirements on compute-intensive models and people become more aware of the carbon footprint of training models of increasing levels of complexity. is.
“Physical manufacturing and testing of biological designs is currently a major bottleneck in biotechnology. As such, this is an important step in the development of biological designs for new applications, from pharmaceuticals to food to environmental applications. This is very encouraging for the prospect of rapidly prototyping parts.”
