DeepMind launches AlphaFold 3 with molecular interaction prediction
Google DeepMind announced on May 8, 2024, the launch of AlphaFold 3, the latest version of its artificial intelligence system for protein structure prediction. Unlike previous versions, AlphaFold 3 can model not only proteins but also their interactions with DNA, RNA, and other molecules — a significant advance for computational biology and drug discovery.
What's new?
While the original AlphaFold (released in 2020) revolutionized protein structure prediction, AlphaFold 3 goes further by simulating how molecules bind to biological targets. This ability to predict molecular interactions is crucial for understanding cellular processes and developing new drugs.
Free access for academic research
The tool is freely available to academic researchers, who can access it via a server maintained by the lab. DeepMind hopes open access will accelerate discoveries in areas such as neglected diseases, structural biology, and protein engineering.
Impact on drug discovery
AlphaFold 3 can dramatically reduce the time needed to identify drug targets and test drug candidates in silico. The ability to simulate interactions between proteins and potential drugs allows researchers to prioritize the most promising compounds before costly and time-consuming lab tests.
Next steps
DeepMind intends to continue improving the model and expanding its database. Although the current version already represents a qualitative leap, the company envisions future iterations capable of modeling even more sophisticated macromolecular complexes, such as ribosomes and transcription complexes.