{"id":1505,"date":"2020-12-01T18:03:48","date_gmt":"2020-12-01T18:03:48","guid":{"rendered":"https:\/\/thenextweb.com\/?p=1330021"},"modified":"2020-12-01T18:03:48","modified_gmt":"2020-12-01T18:03:48","slug":"how-deepminds-protein-folding-breakthrough-could-transform-drug-development","status":"publish","type":"post","link":"https:\/\/www.londonchiropracter.com\/?p=1505","title":{"rendered":"How DeepMind\u2019s protein-folding breakthrough could transform drug development"},"content":{"rendered":"\n
<\/div>\n

An AI system developed by Google\u2018s DeepMind has made a \u201conce in a generation\u201d breakthrough that could have a dramatic impact on the way we treat diseases: accurately predicting how proteins fold into shapes.<\/p>\n

Proteins underpin almost every biological process. Each one is comprised of strings of amino acids that fold into countless elaborate shapes that determine what they do.<\/p>\n

Predicting these structures can help us understand how proteins cause diseases \u2014 and develop drugs to treat them. But scientists currently know the exact 3D shapes of just a fraction of the 200 million-plus proteins they\u2019ve identified.<\/p>\n

[Read: How to build a search engine for criminal data<\/a>]<\/em><\/p>\n

Figuring out a protein\u2019s precise form is typically a time-consuming and expensive process. But DeepMind says its AlphaFold system can predict their structures in a matter of days.<\/p>\n

DeepMind tested its performance at the Critical Assessment of protein Structure Prediction (CASP) competition, a biannual experiment in predicting the 3D structure of proteins.<\/span><\/p>\n

The London-based lab had won the 2018 edition of the challenge, but its predictions lacked the accuracy required to be biologically useful. This year, DeepMind revamped the system.<\/p>\n

The latest version of AlphaFold uses a neural network system to identify<\/span> the structure of a folded protein as a \u201cspatial graph,\u201d where residues are the nodes and edges connect the residues. <\/span><\/p>\n

The system was trained on around 170,000 protein structures from the Protein Data Bank, and uses a relatively modest amount of computing power \u2014 roughly equivalent to 100-200 GPUs run over a few weeks.<\/p>\n