View source for Foldit

{{Short description|Online puzzle game in which players fold proteins, developed at the University of Washington}}
{{Infobox video game
| title = Foldit
| developer = University of Washington (Department of Computer Science and Engineering; Department of Biochemistry)
| platforms = Windows, macOS, Linux
| released = 2008
| genre = Puzzle, citizen science
| modes = Multiplayer (solo and groups)
}}
'''Foldit''' is an online puzzle game in which players fold proteins on screen. It was developed at the University of Washington and released in 2008. Players pull, twist and shake a chain of amino acids with tools taken from the Rosetta software suite, and the game scores each arrangement by its computed energy, so that the lowest score wins. The best solutions have been used by crystallographers and protein designers. Foldit is one of the best known examples of [[citizen science]] done through a game, and the company [[Molecular Reality Corporation]] has named it among the comparisons for its own games.

== Gameplay ==
In the account published in ''Nature'' in 2010, Foldit "engages non-scientists in solving hard prediction problems". Players "interact with protein structures using direct manipulation tools and user-friendly versions of algorithms from the Rosetta structure prediction methodology, while they compete and collaborate to optimize the computed energy".<ref name="cooper">{{cite journal |last1=Cooper |first1=S. |last2=Khatib |first2=F. |last3=Treuille |first3=A. |last4=Barbero |first4=J. |last5=Lee |first5=J. |last6=Beenen |first6=M. |last7=Leaver-Fay |first7=A. |last8=Baker |first8=D. |last9=Popović |first9=Z. |last10=Foldit players |title=Predicting protein structures with a multiplayer online game |journal=Nature |volume=466 |issue=7307 |pages=756-760 |year=2010 |doi=10.1038/nature09304}}</ref> The paper lists "Foldit players" among its authors. Top-ranked players were found to excel at refinement problems that need large rearrangements of the backbone to bury hydrophobic residues, and players working together developed new strategies and algorithms, which, the authors noted, meant that players explore the space of search strategies as well as the space of shapes.

Players can also write their folding strategies down as "recipes", share them and change them. In the year after those tools were added to the game, players developed more than 5,400 recipes by writing new ones and recombining others. The most successful ones spread through the player population, and two became dominant. When the researchers examined the two, they found a striking similarity to an unpublished algorithm that scientists had developed over the same period, and benchmark calculations showed that the algorithm, found independently by scientists and by players, outperformed previously published methods.<ref name="recipes">{{cite journal |last1=Khatib |first1=F. |last2=Cooper |first2=S. |last3=Tyka |first3=M. D. |last4=Xu |first4=K. |last5=et al. |title=Algorithm discovery by protein folding game players |journal=Proceedings of the National Academy of Sciences |volume=108 |issue=47 |pages=18949-18953 |year=2011 |doi=10.1073/pnas.1115898108}}</ref>

== Results ==
A paper of September 2011 reported that players had helped solve a structure that crystallographers had been unable to solve by molecular replacement: the protease of the Mason-Pfizer monkey virus, an enzyme of a retrovirus of the family that includes HIV. After a wide range of attempts had failed, the researchers gave players the problem of building accurate models of the protein, and the models were good enough for molecular replacement to succeed. The paper lists two player groups, the Foldit Contenders Group and the Foldit Void Crushers Group, among its authors.<ref name="khatib">{{cite journal |last1=Foldit Contenders Group |last2=Khatib |first2=F. |last3=Foldit Void Crushers Group |last4=DiMaio |first4=F. |last5=Cooper |first5=S. |last6=et al. |title=Crystal structure of a monomeric retroviral protease solved by protein folding game players |journal=Nature Structural & Molecular Biology |volume=18 |issue=10 |pages=1175-1177 |year=2011 |doi=10.1038/nsmb.2119}}</ref>

In 2012 players were asked to remodel the backbone of a computationally designed enzyme, a Diels-Alderase. After several iterations of design and testing, a helix-turn-helix motif of 24 residues, including an insertion of 13, raised the activity of the enzyme more than 18-fold, and X-ray crystallography showed the insertion in the position of the player model.<ref name="eiben">{{cite journal |last1=Eiben |first1=C. B. |last2=Siegel |first2=J. B. |last3=Bale |first3=J. B. |last4=Cooper |first4=S. |last5=et al. |title=Increased Diels-Alderase activity through backbone remodeling guided by Foldit players |journal=Nature Biotechnology |volume=30 |issue=2 |pages=190-192 |year=2012 |doi=10.1038/nbt.2109}}</ref>

Design followed folding. In the 2019 study players were given a fully extended chain and asked to craft both a folded structure and a sequence that would encode it. One hundred forty-six player designs, with sequences unrelated to those of natural proteins, were encoded in synthetic genes. Fifty-six were expressed and soluble in ''Escherichia coli'' and adopted stable folded structures. They represented 20 folds, including one not seen in natural proteins, and the high-resolution structures of four designs were nearly identical to the player models.<ref name="koepnick">{{cite journal |last1=Koepnick |first1=B. |last2=Flatten |first2=J. |last3=Husain |first3=T. |last4=Ford |first4=A. |last5=et al. |title=De novo protein design by citizen scientists |journal=Nature |volume=570 |issue=7761 |pages=390-394 |year=2019 |doi=10.1038/s41586-019-1274-4}}</ref> David Baker, an author of several of these papers, shared the 2024 Nobel Prize in Chemistry for computational protein design, with Demis Hassabis and John Jumper, who were honored for protein structure prediction (see [[AlphaFold]]).<ref name="nobel">{{cite web |url=https://www.nobelprize.org/prizes/chemistry/2024/summary/ |title=The Nobel Prize in Chemistry 2024 |publisher=Nobel Prize Outreach |access-date=October 25, 2026}}</ref>

== At Molecular Reality ==
The company has set its games beside Foldit twice in its public text. A list of answers for visitors, written in 2024, took the question of how its games compare to "the Borderlands game/Folditathome/EteRNA and other gamified citizen science projects" and replied that [[Maxine's Quest]] "is a way to sweeten the drudgery of collecting data with purpose-built devices".<ref name="records">{{cite web |title=Company records |publisher=Molecular Reality Corporation |date=October 2026}}</ref> And in [[The Disclosure Crawl]], the company's quiz dungeon, the correct answer to a boss's charge that "Citizen science is GARBAGE IN, AI-HALLUCINATED GARBAGE OUT!" begins "Galaxy Zoo, Foldit, and eBird proved citizen science produces publishable results".<ref name="crawl7">{{cite web |url=https://molecularreality.com/static/dandd-game/level-07-dragons-lair.html |title=Dungeons & Diligence: The Disclosure Crawl, Level 7: The Dragon's Lair |publisher=Molecular Reality Corporation |date=2026-04}}</ref> A mock magazine article of November 2025 that the company published under the name of the fictional journalist [[Hypatia Morrison]] names Foldit, with Galaxy Zoo and [[EteRNA]], as a precedent for research done by players.

Foldit's players work on models in a simulation. The company's term for a game whose input drives a physical instrument is [[experigaming]], and its players are called [[Player Scientist|Player Scientists]].

== See also ==
* [[EteRNA]]
* [[Borderlands Science]]
* [[Protein Data Bank]]

== References ==
{{Reflist}}

{{Portal bar|Molecular Reality}}

[[Category:Citizen science]]
[[Category:Protein folding]]
[[Category:Video games developed in the United States]]
[[Category:Puzzle video games]]
[[Category:Biology in video games]]
{{Video-game-stub}}