Protein structure is the core of biochemistry and has profound implications for medicine and technology. With thousands of protein structures being established, there will be improvement in the productivity of pharmaceutical research pipelines. Google‘s AlphaFold 2 indisputably won the 14th Critical Assessment of Structural Prediction competition, a biannual blind test where computational biologists try to predict the structure of several proteins whose structure has been determined experimentally — yet not publicly released. Their results are so incredibly accurate that many have hailed this code as the solution to the long-standing protein structure prediction problems.
DeepMind—a British subsidiary of Alphabet Inc.(a.k.a. Google)—announced it’d solved a “grand challenge” in biology. The challenge is known as “the protein folding problem.” It required the AI company to develop a way to predict proteins’ structures based solely on their amino acid sequences. Using the cutting-edge method, AlphaFold, the company says it will sequence every protein scientists know. And it’s already released structures for 350,000 of them on a searchable database.
MIT Technology Review reported on DeepMind’s newest application of AlphaFold. Notably, in a very Google-ish fashion, the company has set up a searchable database that anyone can use online. AlphaFold DB’s tool is already up and running, and people can search the entire human proteome. As well as the proteomes of 20 other scientifically relevant organisms such as yeast, fruit flies, and mice.
AlphaFold can structure the various proteins that make up an organism’s proteome. Using machine learning, the organism’s entire set of constituent proteins. DeepMind trained the AI on roughly 170,000 protein structures from a large database. The training allowed the AI to “learn” which amino acid sequences lead to which types of proteins. Thanks to its gleaning patterns from the data, AlphaFold can now identify novel protein structures after only seeing their amino acid sequences.