DeepMind AI Unlocks Structures for Every Known Protein
DeepMind uses its AlphaFold AI to predict the shapes of hundreds of thousands of proteins and plans to map over 100 million more in the coming months. This massive database gives scientists instant access to crucial biological structures that previously took months to uncover in the lab.
DeepMind announces a major leap forward for biology by using its AlphaFold AI to predict the three-dimensional structures of 350,000 proteins, including almost the entire human proteome. The company plans to expand this database to over 100 million protein structures in the next few months, effectively mapping every protein known to science. This massive undertaking represents the largest and most impactful project DeepMind has ever undertaken.
Understanding the specific shape of a protein is essential for scientists because it reveals the protein's function and role in the body. Traditionally, researchers spend weeks or months in a laboratory to determine a single protein structure, but AlphaFold completes this task with atomic accuracy in just a day or two. By making these predicted shapes freely available, DeepMind gives researchers a powerful tool to understand how diseases work and to develop new treatments.
The creation of this comprehensive database eliminates the need for individual labs to run the complex AlphaFold software on their own computers. Instead, biologists simply look up the structures they need, saving immense amounts of time and computational resources. DeepMind cofounder Demis Hassabis believes this technology ultimately has the potential to create massive real-world impacts far beyond the field of artificial intelligence.