The European Molecular Biology Laboratory (EMBL) and DeepMind have partnered to create a database of predicted structures of proteins for the human proteome of about 20,000 genes using artificial intelligence. The new technology could accelerate the discovery of drugs to treat diseases. The team has published the Highly Accurate Protein Structure Prediction for the Human Proteome study in the Nature journal.
The AI enables scientists to look into the 3D structure of proteins. It also provides a lot of data that could make room for advances in the treatment of diseases and the development of new drugs
AlphaFold works better than traditional methods for predicting protein structure
According to John McGeehan, a structural biologist professor from the University of Portsmouth, traditional methods for predicting protein structure take a lot of time and money. However, this new AI technology could make the process faster and more cost-effective.
Dr. Demis Hassabis, the co-founder and CEO of DeepMind, adds that they have created a more accurate prediction of the human genome using the technology named AlphaFold. AlphaFold made precise and complete predictions for about 58% of the amino acids in a human proteome, predicting 35.7% of amino acids with a high confidence degree.
Traditional methods for predicting protein stricture could take up to six months. However, AlphaFold only takes a few minutes. Thus, the technology could speed up the discovery of medication. It could also make enzymes to break down plastics or crops that withstand climate change. These are projects the scientists were working on at the time. As a result of the discovery, they will be able to work faster on these projects.
The Deputy Director-General and EMBL-EBI Director, Ewan Birney, adds that the dataset could become the most critical development since mapping the human genome. It could also complement other technologies to make more discoveries that could help scientists understand the world better.
Scientists could use the AlphaFold database with other organisms
The database predicts not only the structure of the human proteome but also proteomes from other organisms. These organisms include tuberculosis bacteria, malaria parasites, zebrafish, mice, fruit fly, and E.Coli.
