Foldit is a citizen science puzzle game that challenges players to solve real-world protein-folding problems, leading to disease research and drug development discoveries.
In 2008, the University of Washington Biochemistry and the UW Department of Biochemistry released Foldit, a crowd-sourcing protein-folding puzzle game. Players predict the 3D structures of the proteins, seeking to uncover the most stable and energy-efficient states. Foldit was among the first citizen science games (CSGs) designed for real-world scientific problem-solving, demonstrating that gameplay could facilitate meaningful public participation in research. By the early 2010s, the term Games With a Purpose (GWAP) was often used to describe such efforts (Miller et al., 2023).
The origin
Resolving the shape of proteins can lead to drug discoveries and cures for human diseases since the shape of the protein defines its function and how it interacts with other molecules. It all started with Rosetta@home, a protein structure prediction programme created by molecular biologist David Baker. Rosetta's "network is capable of 100 trillion calculations per second, dwarfing most supercomputers" (Bohannon, 2010), circulating calculations through thousands of volunteer home computers worldwide. Some Rosetta users said they occasionally saw potential solutions but felt frustrated when they could not make any corrections. Baker then approached computer science professors David Salesin and Zoran Popović to conceptualise and create an interactive program, and they turned it into a video game instead (Technology Review, 2008).
Determining protein structures can lead to drug discoveries and cures for human diseases since a protein's shape defines its function and interactions with other molecules. Foldit's origins trace back to Rosetta@home, a protein structure prediction program created by molecular biologist David Baker. Rosetta's network performs 100 trillion calculations per second, surpassing most supercomputers, distributing tasks across thousands of volunteer home computers worldwide (Bohannon, 2010). Some Rosetta users reported seeing potential solutions but felt frustrated that they could not make direct modifications. Recognising this limitation, Baker then approached computer science professors David Salesin and Zoran Popović to develop an interactive solution, and they turned it into a video game instead (Technology Review, 2008).
The human vs. machine
In 2019, a study compared the performance of Foldit players with Rosetta's automated strategies. The findings revealed that Foldit players significantly outperformed automated approaches. As shown in Fig. 2a, Foldit players (red, blue, and green) explored new regions with significant energy increases, whereas Rosetta's strategies remained limited and systemic. Red circles correspond to snapshots of the trajectory displayed in Fig. 2b. Finally, Fig. 2c illustrates the strategic trajectory of Foldit players; each colour represents different cooperating Foldit players, with the final structure (marked by a star) achieved after 17 branch points (Koepnick et al., 2019).
A decade-long problem
Scientists hoped that gamers manipulating protein structures might help uncover the enzyme M-PMV Retroviral protease, which is crucial for understanding how HIV multiplies. After scientists and computer programs failed to determine the structure for over a decade, the M-PMV Foldit puzzle was created. In just three weeks, gamers produced a 3D model of the enzyme, accurate enough for molecular replacement (Khatib et al., 2011; Zoran, 2011). This discovery paved the way for developing antiretroviral drugs, including treatments for HIV (Khatib et al., 2011).
If a protein researcher is struggling with a particular problem, they will create a Foldit puzzle for their problem. — Foldit website.
Foldit has hundreds of thousands of registered players who contribute to cancer and Alzheimer's research, among other causes. Gameplay has since evolved to allow players to design never-before-seen synthetic proteins, which led David Baker to experiment with protein design.
Foldit Legacy
If we can mimic the pinnacle of intuition in Go, then why couldn't we map that across to proteins? —Demis Hassabis, co-founder of DeepMind (Heaven, 2022).
Simultaneously, Demis Hassabis, co-founder of DeepMind, started experimenting with game systems and AI. After experimenting with an automated game system for old arcade classics, he took on a more complex game system and created AlphaGo, which plays the ancient strategy board game Go. AlphaGo eventually defeated the Go (board game) world champion, Lee Sedol. Seeking a new challenge, Hassabis looked at Foldit's success, which led him to "think that AI could maybe try to mimic that intuitive capability that those gamers were demonstrating" (BBC, 2020). Hassabis and scientist John M. Jumper started testing with protein folding, and in 2020, DeepMind released AlphaFold2. The programme's success led experts to claim that the half-century protein folding problem has been broadly solved. In 2024, the Nobel Prize in Chemistry was awarded to David Baker for computational protein design and Demis Hassabis and John M. Jumper for protein structure prediction.
References
Koepnick, B., Flatten, J., Husain, T. et al. (2019). De novo protein design by citizen scientists. Nature, 575, 184–188. https://doi.org/10.1038/s41586-019-1274-4
Khatib, F., DiMaio, F., Foldit Contenders Group, et al. (2011). Crystal structure of a monomeric retroviral protease solved by protein folding game players. Nature Structural & Molecular Biology, 18(10), 1175–1177. https://doi.org/10.1038/nsmb.2119
Bohannon, J. (2010, August 4). Video game helps solve protein structures. Science. Retrieved January 31, 2025, from https://www.science.org/content/article/video-game-helps-solve-protein-structures
Miller, J. A., Vepřek, L. H., Deterding, S., & Cooper, S. (2023). Practical recommendations from a multi-perspective needs and challenges assessment of citizen science games. PLoS ONE, 18(5), e0285367. https://doi.org/10.1371/journal.pone.0285367
Cooper, S., Khatib, F., Treuille, A., et al. (2010). Predicting protein structures with a multiplayer online game. Nature, 466, 756–760. https://doi.org/10.1038/nature09304
Zoran, A. (2011). Crystal structure of a monomeric retroviral protease solved by protein folding game players. https://homes.cs.washington.edu/~zoran/NSMBfoldit-2011.pdf
Heaven, W. D. (2022, February 23). This is the reason Demis Hassabis started DeepMind. MIT Technology Review. Retrieved January 31, 2025, from https://www.technologyreview.com/2022/02/23/1045016/ai-deepmind-demis-hassabis-alphafold
Saplakoglu, Y. (2024, June 26). How AI revolutionized protein science—but didn't end it. Quanta Magazine. Retrieved January 31, 2025, from https://www.quantamagazine.org/how-ai-revolutionized-protein-science-but-didnt-end-it-20240626/
BBC News. (2020, December 2). DeepMind co-founder: Gaming inspired AI breakthrough. BBC News. https://www.bbc.co.uk/news/technology-55157940
Fold.it. (n.d.). Science. Retrieved January 31, 2025, from https://fold.it/science
Rosetta@home. (n.d.). What is Rosetta@home? Retrieved January 31, 2025, from https://boinc.bakerlab.org/rosetta/rah/rah_about.php
Tiz Creel of Living Things Studio ©2025
Thank you for reading 🫀
Keep it playful.
Komentarze