Written and maintained by CASRAI Editorial Board
Last updated
Short answer: Foldit is a free online puzzle game, run by university research groups, in which players manipulate protein structures to lower a computed energy score. Published studies show that players’ solutions and the strategies they wrote (“recipes”) have fed back into protein-structure and enzyme-design research.
Definition and scope
Foldit describes itself as “a revolutionary crowdsourcing computer game enabling you to contribute to scientific research” (fold.it, read 2026-11-08). Its listed developers are the Center for Game Science and the Institute for Protein Design at the University of Washington, together with groups at Northeastern, UMass Dartmouth, UC Davis, Vanderbilt and the University of Denver. The site lists DARPA, NSF, NIH and HHMI among its supporters.
In the first peer-reviewed description, Cooper et al. (Nature 2010, 466:756-760, doi:10.1038/nature09304) characterise Foldit as a multiplayer online game that engages non-scientists in hard protein structure prediction. It is an example of citizen science in which the volunteer task is a design and optimisation puzzle rather than data labelling.
How it works in practice
According to the 2010 paper, players use direct-manipulation tools and user-friendly versions of Rosetta algorithms to reshape a protein, and compete and collaborate to optimise the computed energy of the structure. The authors report that top players were particularly good at refinement that needs substantial backbone rearrangement, for example to bury hydrophobic residues.
Players also changed the method, not only the answers. Khatib et al. (PNAS 2011, 108:18949-18953, doi:10.1073/pnas.1115898108) describe how players encode strategies as “recipes”. In the year after the recipe tools were introduced, players produced over 5,400 different recipes, and two became dominant. One resembled an algorithm a scientist had developed independently in the same period, and that algorithm outperformed previously published methods in the authors’ benchmark. The paper concludes that Foldit player solutions can in some cases outperform state-of-the-art computational methods.
A later report moved from prediction to design. In Nature Biotechnology (2012, 30:190-192, PMID 22267011), players were asked to remodel the backbone of a computationally designed Diels-Alderase. Several iterations of design and laboratory characterisation produced a 24-residue helix-turn-helix motif, including a 13-residue insertion, that increased enzyme activity more than 18-fold, and X-ray crystallography showed the insertion adopting the structure positioned as in the Foldit model.
Three kinds of contribution compared
| Contribution | What the player supplied | Published evidence |
|---|---|---|
| Structure refinement | Manual backbone moves that lower the computed energy | Cooper et al., Nature 2010 |
| Strategy (“recipe”) development | Reusable procedures; over 5,400 in one year, two dominant | Khatib et al., PNAS 2011 |
| Enzyme redesign | A remodelled backbone segment tested in the laboratory | Nature Biotechnology 2012 (abstract) |
Where it matters for labs and research offices
For a lab or research office, Foldit is a concrete case of how crowd contributions are recorded. In the PubMed records for the 2010 and 2011 papers, the final author listed is “Players”, so the contributors appear as a group on the author line rather than as named individuals. How a group like this is treated under journal rules is the subject of the dictionary entries on group authorship and citizen-scientist contribution.
When you cite or build on such work, record the paper and its DOI, the game version or tool you relied on if the paper names one, and whether the result was computational only or confirmed by laboratory measurement. The enzyme study is the confirmed case in the table above; the first two rows are computational results.
The question of who counts as an author versus a person named in an acknowledgments section is decided by each journal’s policy, so check the target journal’s own instructions rather than assuming the Foldit convention applies.
Common misunderstandings
Is Foldit just a game with no research use?
No. The three papers above are peer-reviewed reports in which player activity contributed to research results.
Does it replace computational biologists?
The sources do not say that. The 2010 paper describes players working with Rosetta-derived tools, and the 2011 paper compares player recipes with scientist-written algorithms.
Is it the same as crowdsourced replication?
No. Foldit crowdsources a design task; crowdsourced replication distributes re-running of existing studies.
FAQ
What is Foldit?
An online protein-folding puzzle game developed by university research groups in which players optimise protein structures for a computed energy score (fold.it, read 2026-11-08).
Who runs Foldit?
The site lists the University of Washington Center for Game Science and Institute for Protein Design, Northeastern, UMass Dartmouth, UC Davis, Vanderbilt and Denver research groups as developers.
Has Foldit produced published research?
Yes: Nature (2010), PNAS (2011) and Nature Biotechnology (2012) papers are listed under Sources.
Sources
- Foldit, fold.it, read 2026-11-08.
- Cooper S, et al. Predicting protein structures with a multiplayer online game. Nature 2010;466(7307):756-760. doi:10.1038/nature09304, PMID 20686574 (PubMed record read 2026-11-08).
- Khatib F, et al. Algorithm discovery by protein folding game players. PNAS 2011;108(47):18949-18953. doi:10.1073/pnas.1115898108, PMID 22065763 (PMC3223433).
- Increased Diels-Alderase activity through backbone remodeling guided by Foldit players. Nat Biotechnol 2012;30(2):190-192. PMID 22267011 (PubMed record read 2026-11-08).
