Google DeepMind has made a major breakthrough in biology by using AI able to accurately predict the ways proteins fold.
The protein folding problem refers to the high level of complexity involved in the way proteins go from being simple strings of amino acids to folded 3D structures inside cells.
Co-founder of the biennial Critical Assessment of protein Structure Prediction (CASP) experiment, Professor John Moult, said he was happy to see DeepMind offer a solution to this long-standing biological challenge.
“We have been stuck on this one problem – how do proteins fold up – for nearly 50 years,” Professor Moult said.
“To see DeepMind produce a solution for this, having worked personally on this problem for so long and after so many stops and starts, wondering if we’d ever get there, is a very special moment.”
DeepMind solved the notoriously difficult problem using a machine learning system called AlphaFold which has been trained on around 170,000 protein structures and other large databases of protein sequences.
Training the model took “a few weeks” running on 128 cores of Google’s third generation tensor processing unit (TPU), according to the AlphaFold team.
The AlphaFold system can take an amino acid sequence and output a highly accurate folded protein.
The breakthrough could have massive implications for biological research and – especially in the development of new drugs and medical treatments.
“When DeepMind started a decade ago, we hoped that one day AI breakthroughs would help serve as a platform to advance our understanding of fundamental scientific problems,” the AlphaFold team said in a blog post.
“Now, after four years of effort building AlphaFold, we’re starting to see that vision realised, with implications for areas like drug design and environmental sustainability.”
Director of the Max Planck Institute for Developmental Biology, Professor Andrei Lupas, said Deepmind’s research outputs have already shown academic benefits.
“AlphaFold’s astonishingly accurate models have allowed us to solve a protein structure we were stuck on for close to a decade, relaunching our effort to understand how signals are transmitted across cell membranes,” she said.
DeepMind became Google’s AI research arm after the tech giant acquired it in 2014.
Research from DeepMind made global headlines in 2017 after its self-taught Go-playing AI beat the best player in the world at the ancient board game.