[tweetmeme source=”scwai” only_single=false]
Was reading our latest issue of the The Economist over the weekend and was delighted by the following article: “Scientific discovery games: Game not over“. Seth Cooper and colleagues at the University of Washington have designed an online video game called Foldit that harnesses the puzzle-solving skills of its players to predict protein structure. They’ve published their findings in Nature (PubMed entry, Nature abstract). Ed Yong has a great write-up of the paper here.
The rules of protein folding really lend itself well to the context of a video game. Proteins are made up of chains of amino acids folded into 3D shapes. There are 20 different amino acids, all with their own unique properties. For example, there is one called glycine that allows a large amount of floppiness or flexibility in the amino acid chain. Another called proline does the opposite, forcing a fixed kink in the chain. There are ones like tryptophan that have a large side group sticking out and so get in the way of other amino acids when the protein starts to fold. When arranged in a particular order, small groups of amino acids might take on the form of a spiral (alpha-helix) or something like an accordion fold (beta-sheet). Sounds simple enough to begin with, but the rules get more complicated as the chain grows. Proteins consist of hundreds and thousands of amino acids! One of the proteins I worked on in grad school, Cdc24, consists of 854 amino acids. According to the Wikipedia article on proteins, the largest known proteins are the titins, which have almost 27,000 amino acids (hence the name).
Rosetta is a set of software based on these amino acid rules, allowing users to use computing power to sort through all these rules and predict protein structure. The actions in Foldit are based on the ones used in Rosetta. One of the most interesting findings in this study by Cooper and colleagues is that the Rosetta program and the game players excelled at different parts of the protein-folding puzzle. This suggests that we might optimize protein structure prediction by combining computer and human ingenuity. The team is using the findings from Foldit to improve the Rosetta software.
I’m not a gamer – I don’t have the patience. But plenty of my friends are happy to spend hours playing games that appear to utilize some degree of strategy and puzzle-solving. Maybe after they’ve beat their latest shooting and colony-building games, they can give Foldit a try. Some extra incentive? “Players F.” in the citation below refers to Foldit players. A Nature authorship (even if anonymous) is a pretty good prize for a video game.
Full citation: Cooper S, Khatib F, Treuille A, Barbero J, Lee J, Beenen M, Leaver-Fay A, Baker D, Popović Z, Players F. Predicting protein structures with a multiplayer online game. Nature. 2010 Aug 5;466(7307):756-60.
Side note: Most of what I read these days is online, stumbled upon on my own or via links in my Twitter feed. Thus, it’s notable that I came across this article first in its print form. We still subscribe to The Economist.