AlphaFold AI System can Change the World

AlphaFold AI System can change the world

While 2020 was a difficult time across many metrics, it was an exciting year for protein science!

AlphaFold gained international acclaim for advancing protein structure prediction after blowing the competition out of the water at CASP in both the 2018 and 2020 competitions (1). The AlphaFold system uses AI to decipher an unknown protein structure by predicting the distance and bond angle between pairs of amino acids.  Now that the technology exists, leaders in research and industry are brainstorming new ways to use protein structure prediction. The following sections explore different uses of AlphaFold and discuss the societal implications of protein structure prediction.

 

Improving health and drug treatments

The medical field is the most obvious area where AlphaFold can accelerate progress. Increasing precision and speed for drug design has the potential to make drugs more effective, more available, and less expensive than ever before. Increasing the number of known human proteins opens up new therapeutic avenues for diseases like cancer, Alzeihmer’s, and Parkinson’s (2). AlphaFold has also been deployed in the global fight against COVID-19. Protein structures are instrumental in determining how viruses infect cells and how the immune cell recognizes viral invaders and eventually fights them off. The AlphaFold team released hypothesized structures of several viral proteins to improve understanding of how a the virus functions and serve as a starting point for new experimental therapies (3).

 

Advancing sustainability goals

In the near future, AlphaFold can also be used to improve environmental sustainability and address the challenges of climate change. For example, finding new proteins that can degrade plastic or capture carbon may help to lessen the effects of industrial pollution and global warming (1). Protein structure prediction could also be of great economic value for designing genetically modified crops that are drought and pest resistant, new synthetic materials, cosmetics, and natural products. DeepMind’s commitment to keeping AlphaFold code open source means that more people will have access to use the technology in new and innovative ways.

 

What are the risks?

Without careful oversight, AlphaFold’s technology could also cause harm. Misuse (or disuse) of protein engineering could promote development of new toxins that are more stable and more lethal than a toxin that occurs naturally (4). In a similar way to designing more effective drugs, knowing the structure of the cellular target, means that more efficient bioweapons can be designed as well.

AlphaFold has made significant progress in solving a  50-year biological puzzle, and the discovery will have positive impacts across diverse fields of research and development. But with great power comes great responsibility, and strong guidance will be needed to ensure that the advantages of this powerful new technology are not overshadowed by the costs.

The next and final installment of our three-part series will follow the story of Team AlphaFold as they eagerly joined the fight against COVID-19.

 

References

  1. AlphaFold Team. (2020). AlphaFold: a solution to a 50-year-old grand challenge in biology. Retrieved from https://deepmind.com/blog/article/alphafold-a-solution-to-a-50-year-old-grand-challenge-in-biology
  2. ODSC Team. (2020). DeepMind’s AlphaFold AI could Accelerate Drug Discovery. Retrieved from https://opendatascience.com/deepminds-alphafold-ai-could-accelerate-drug-discovery/
  3. Jumper, J., Tunyasuvunakool, K., Kohli, P., Hassabis, D., AlphaFold Team. (2020). Computational predictions of protein structures associated with COVID-19. Retrieved from https://deepmind.com/research/open-source/computational-predictions-of-protein-structures-associated-with-COVID-19
  4. Tucker, J. B., & Hooper, C. (2006). Protein engineering: security implications. The increasing ability to manipulate protein toxins for hostile purposes has prompted calls for regulation. EMBO reports, 7 Spec No(Spec No), S14–S17. https://doi.org/10.1038/sj.embor.7400677