Data Science Investment Counter —Funding raised by UK data science companies in 2018.
£ 5.640 Million

Google’s DeepMind algorithm wins protein-folding contest

DeepMind’s latest AI program, AlphaFold, has proven to be the best software existing at a predicting the 3D shapes of proteins, organisers of an international conference in Cancun reported on Sunday.

DeepMind’s algorithm has beaten 98 other programs, the Guardian reported. In one part of the contest, its software accurately predicted the structure of 25 out of 43 proteins, while the second-place algorithm only got three of the 43.

The shape of proteins is fundamental in the understanding of many biological processes, and being able to predict their shapes could prove a key step in finding molecules that may be useful in the creation of new medicines for long-term diseases such as diabetes, Parkinson’s and Alzheimer’s disease.

It was the first time that DeepMind, best known for beating World Go Champion Ke Jie last year, had entered software in the protein-folding competition, run by the Protein Structure Prediction Center.

The techniques used to train the algorithm to predict protein shapes were not identical to the ones DeepMind had used to master Go,  the company said in a blog post on Sunday, but there were some similarities.

Just like the Go-playing algorithm did not need human interaction to learn, the folding software was programmed to generate protein shapes from scratch, without having access to known examples of protein shapes to use as models. Two different deep neural networks were used to arrive at the protein-shape predictions.

“The 3-D models of proteins that AlphaFold generates are far more accurate than any that have come before — making significant progress on one of the core challenges in biology”, the post said.

The company said they spent two years developing the software. The name links back to AlphaGo, its original champion Go software.

DeepMind has often said it wants to apply its AI techniques to fundamental scientific problems. Demis Hassabis, the company’s co-founder and chief executive officer, had previously mentioned protein-folding and finding new potential targets for drug development as areas the company was exploring.

“For us, this is a really key moment,” said Demis Hassabis, co-founder and CEO of DeepMind. “This is a lighthouse project, our first major investment in terms of people and resources into a fundamental, very important, real-world scientific problem.”

“It’s never been about cracking Go or Atari,” Hassabis said, “it’s about developing algorithms for problems exactly like protein folding.”


Co-working space and blog dedicated to all things data science.

Subscribe to our newsletter