Google has reached an important milestone in the development of artificial intelligence. DeepMind researchers presented AlphaFold 3, a new model capable of predicting the structure and interactions of all life molecules. AI promises to revolutionize the understanding of biological processes and pave the way for drug discovery.
In an article published in the magazine NatureThe scientists describe in detail the development and capabilities of their new model. AlphaFold 3 can not only decipher the 3D structures of proteins, but also predicts how they interact with a wider range of molecules. This includes DNA, RNA and small molecules such as drugs.
Alpha fold 3 can generate 3D structure from a list of molecules, showing how they all fit together. The model is also capable of simulating chemical modifications of these molecules that control healthy cell function and that, if altered, can cause disease.
This is possible thanks to the next generation architecture based on dataset covering all biological molecules. AlphaFold 3 uses an improved version of the Evoformer module, the neural network block that underpins its predecessor. After AlphaFold 3 molecular data processing uses a diffusion network similar to that used in artificial intelligence imaging.to refine your predictions.
A complete map of molecular interactions not only opens up a deeper understanding of cellular machinery. According to Google, AlphaFold 3 could benefit materials development, accelerate the development of new drugs, and improve genomic research..
AlphaFold 3 AI will help discover new drugs
DeepMind researchers claim that AlphaFold 3 outperforms existing methods in ability to predict molecular interactions. This model is the first artificial intelligence system to outperform physical tools in predicting biomolecular structures.
This translates to significant advances in drug discovery, in particular to predict how drugs and antibodies bind to proteins, thereby influencing their impact on human health and disease. “AlphaFold 3 achieves unprecedented accuracy in predicting drug interactions, including protein binding to ligands and antibodies to target proteins,” the scientists added.
Some researchers suggest that drugs discovered using artificial intelligence have higher success rates in the early stages of testing. However, the molecules proposed by AlphaFold 3 will require validation and clinical trials.
“AlphaFold 3 brings the biological world into high definition. It allows scientists to see cellular systems in all their complexity, through structures, interactions and modifications. This new window into the molecules of life shows how they are connected and helps us understand “how these connections influence biological functions such as drug action, hormone production and the DNA repair process that keeps us healthy.”
The potential of this model extends beyond drug development. AlphaFold 3 will be an important step in developing biorenewable materials or creating more sustainable crops. In one of its predictions, artificial intelligence modeled the composition of the enzyme present in Verticillium dahliaa fungus that causes discoloration and death of many plant species.
AlphaFold 3 will be available through the DeepMind server, which will be free for academic and non-commercial use.
Source: Hiper Textual
