22 May, 12 | by BMJ Group
“You need a lot of patience to play this game,” one of the Foldit community replied after I asked for some tips on how to rebuild my protein’s backbone. “Just keep wiggling,” another suggested. The frustration of watching my score plummet as my protein unfolded was getting to me. I might have given up at this point, but Foldit is not just a geeky thinking man’s Angry Birds. By playing I’m also contributing to science, joining the effort to better understand protein structures that may one day lead to new cures for disease. With that goal in mind, I start again, attach my virtual rubber bands to different points on the protein, and click, wiggle, and cross my fingers.
Figuring out the atomic structure of proteins is one of biochemistry’s biggest challenges—an important goal, since understanding protein structure enables targets for new drugs to be developed. Given that they are usually formed of hundreds (or even thousands) of amino acids, there are innumerable ways in which they can fold up. Even the most powerful computer programmes have difficulty working out their actual structures. Foldit’s makers (a collaboration between the Centre for Game Science at the University of Washington and the University of Washington Department of Biochemistry), came up with a novel solution to this problem: make a computer game and get the public to do it. Their results so far are promising: so could this strategy, known as crowdsourcing, be the future of scientific and medical research?
Foldit was borne out of Rosetta@home, a project that uses spare computer processing power—in the form of a screensaver—to work out how proteins fold. In 2008, when Foldit launched, gamers were given the chance to do this themselves. Since then the researchers behind Foldit have been gathering evidence to demonstrate that it works. “Gamers come up with different classes of solutions which in some cases are better than automated methods,” says David Baker, Professor of Biochemistry at the University of Washington, and one of Foldit’s principal investigators. “An area where it is easy to see why humans could be better than completely automated methods is in protein design.” Their recent paper in Nature Biotechnology describes how Foldit players came up with the first “crowdsourced” redesign of a protein (1). Players were asked to come up with a new structural element to improve an active site on an enzyme. What they came up with was found to have an 18 fold increased activity than the original enzyme.
The game itself, free to download from http://fold.it/, is slick and user-friendly. There is a steep learning curve—perhaps unsurprisingly, given the goal of the game—but once you’ve grasped the basic concepts the game is fun and rewarding to play. The game consists of puzzles and contests where your goal is to manipulate the shape of a given protein to make its structure more stable. The more stable it is, the higher your score. The key to doing this is to pack the protein, making it as small as possible, whilst avoiding letting your side-chains bump into one another. There are various tools that help you to reshape your proteins: the simplest, most effective and, perhaps most bemusing, is to “wiggle” your back-bone (otherwise known as gradient based energy minimisation).
Current puzzles allow you to try your hand at solving the native confirmation of HIV protease, or perhaps model the anthrax protein. You can publish your methods as recipes, so that others can try to improve on it. Foldit’s researchers have found that when players collaborate in this way, their solutions are similar to those developed independently by scientists (2).
So should the medical and pharmaceutical world be following Foldit’s lead by developing other ways to utilise the brains of the millions of computer gamers out there? The idea of crowdsourcing isn’t new: in the nineteenth century the Oxford English Dictionary was compiled thanks to the contributions of thousands of volunteers who sent in their examples of English usage. Amazon’s Mechanical Turk (https://www.mturk.com/), running since 2005, is an online marketplace where companies can outsource so called “human intelligence tasks”—usually simple or repetitive tasks, such as transcribing an audio recording or testing websites. There are plenty of examples of so called citizen science projects within astronomy and conservation (eg www.citizensciencealliance.org/ www.bto.org/volunteer-surveys), but examples within human sciences and medicine are harder to find. Perhaps this is because, in general, medical research doesn’t lend itself to crowdsourcing particularly well. It is easy enough to publish images obtained from a satellite and get the public to look through them for signs of life; developing new drugs usually requires specialist equipment. Where there are opportunities to crowdsource within medical research perhaps researchers are too protective of their ideas and need the courage to take a leap into the unknown. Or maybe they need to be more creative, and take advantage of the untapped resource of the crowd. “There is huge potential for the general public to contribute to science,” says Baker. “There is a lot of brain power out there!”
Tom Nolan is a GP trainee in London.
1. Eiben C, Siegel J, Bale J, Cooper S, Khatib F, Shen B, Foldit Players, Stoddard B, Popovic Z, Baker D. Increased Diels-Alderase activity through backbone remodeling guided by Foldit players. Nature Biotechnology 2012;30:190-192
2. Khatiba F, Cooperb S, Tykaa M, Xub K, Makedonb I, Popovićb Z, Baker D, and Foldit Players. Algorithm discovery by protein folding game players. Proceedings of the National Academy of Sciences 2011;108:18949-18953