On Wednesday Eastern Time, Google DeepMind released a new generation of AlphaFold 3 models for predicting protein structure, which can help scientists more accurately target disease mechanisms and develop more effective therapeutic drugs.
DeepMind researchers state that AlphaFold 3 is an artificial intelligence (AI) model that can predict the structure of biomolecules such as proteins, DNA, RNA, and how they interact.
DeepMind CEO Damis Hassabis said at a press conference on Tuesday that AlphaFold 3 is an important milestone for us. "Biology is a dynamic system, and you must understand how physiological characteristics are generated through the interactions between different molecules in cells. You can see AlphaFold 3 as a big step forward in this direction."
Hassabis added that the breakthrough research paper will be published in Nature on Wednesday, and AlphaFold 3 can significantly reduce the time and funding required to develop life changing treatments.
In addition, DeepMind has launched AlphaFold Server, a free platform for global scientists to use for non-commercial research.
Milestone breakthrough
In 2018, DeepMind launched the first generation AlphaFold model and won first place in the international protein structure prediction competition. In 2020, AlphaFold 2 continued to demonstrate astonishing predictive accuracy and is considered a milestone breakthrough in the field of protein structure prediction.
Now, AlphaFold 3 goes further, predicting the structure of almost all biomolecules and simulating the interactions between these molecules. Although researchers have developed specialized computational methods to simulate interactions between specific types of biomolecules, AlphaFold 3 is the first single system to predict interactions between almost all molecular types with state-of-the-art performance.
If ordinary experimental methods are used to understand the interactions between molecules, it may take several years of research time and the cost is prohibitively high. But if these interactions can be obtained with sufficient accuracy through calculations, biological research can be greatly accelerated.
For example, if researchers believe that a molecule that can bind to a specific protein site may be a promising drug candidate, they can use AI systems like AlphaFold 3 to test potential drug molecules.
Nobel laureate and geneticist Paul Nass commented that AlphaFold is constantly improving and becoming increasingly important for biological research. AlphaFold 3 can predict the structure of complexes between different macromolecules with higher accuracy, as well as the interactions between macromolecules, small molecules, and ions.
Dr. Ivo Tews from the University of Southampton called AlphaFold 3 a leap and stated that his laboratory will use it to develop drugs for treating cancer. He added, "This will save a lot of time and accelerate research by generating models, and then we can explore with new experiments."