DeepMind AI solves 50-year old medical science problem of mapping protein structure

Representative Image: REUTERS/Stefan Wermuth

An artificial intelligence company called DeepMind has built an AI system that found a breakthrough to a grand challenge in Biology which hindered medical advancement and disease fighting mechanisms from progressing.

DeepMind, the company that had become famous for designing computers that could defeat humans at games, has been able to solve a 50-year-old science problem, which will help make huge advancement in the field of biology.

Fifty years ago, scientists had said that a protein’s structure could be predicted by learning its sequence of amino acids. However, it remained a challenge for scientists for all these years because there are literally innumerable ways in which the same protein could fold to take on a 3D structure.

DeepMind’s AI system can swiftly and accurately predict how proteins fold into 3D formations, which scientists were unable to figure until now. According to a report by the Independent, the AI system could drastically change our fight against diseases – ranging from coronavirus to cancer.

DeepMind developed the pathbreaking AI project with the 14th Community Wide Experiment on the Critical Assessment of Techniques for Protein Structure Prediction (CASP14). CASP was formed 25 years ago by a group of scientists to hold international competitions wherein various methods of predicting protein structure would be compared for eventual medical advancement.

Figuring out a protein’s structure or what is commonly referred to as protein folding, used to take years of laborious experimentation, and even so, they were not accurate. However, AlphaFold – the AI system created by Google’s DeepMind, will be able to accurately predict a protein’s structure and map its three-dimensional shapes in only a few hours.

Since the outbreak of the novel coronavirus pandemic, we have been hearing how a “spike in protein” on the virus allows it to bind and invade human cells. Such complex processes among others take place because the amino acids in protein molecules – which are essentially a string of chemical compounds -- fold in intricate ways to create unique shapes that decide how the protein will behave.

Dr John Moult, chair of CASP14, said: “Even tiny rearrangements of these vital molecules can have catastrophic effects on our health, so one of the most efficient ways to understand disease and find new treatments is to study the proteins involved.”