Online game helps predict how RNA folds

A novel hybrid of computer gaming and real-world biochemistry may soon give researchers the ability to predict the complex folding patterns of RNA molecules. This would allow them to synthesise bespoke molecules that can function as chemical sensors or be used in other applications. The folding of RNA molecules is difficult to predict, because each molecule is a long string of units, or bases, that can pair up with each other in many different ways. Because of this, even the best computer algorithms do badly at predicting the shape a molecule will actually take. A team led by computer scientist Adrien Treuille at Carnegie Mellon University in Pittsburgh, Pennsylvania, turned the problem over to online gamers to crack. "Instead of having powerful computers go through the data, why not have lots and lots of people do it?" says Treuille. Players of the game, called EteRNA, are given a target shape and can then join bases in any order. With each change, the computer calculates the most likely shape of the folded RNA molecule, allowing players to adjust bases until they achieve their target. Honing skills EteRNA went public last week, and has already attracted about 5000 players. At first, players will hone their skills by designing simple shapes to hit the game's targets. Within a few months, though, the researchers hope to start them on harder, real-world problems. For example, Treuille hopes to design an RNA molecule that can change shape in the presence of another target molecule, thus acting as a sensor. An earlier game, called FoldIt, which Treuille also helped design, lets players predict the shape of folded proteins. But because RNA molecules are quick and cheap to synthesise, the new game goes a step further. Each week, Treuille's colleague Rhiju Das of Stanford University in California, creates some of the molecules in his laboratory to see whether they really do take the predicted shape. Virtual to real So far, only 4 of about 40 tested have done so. However, the gamers are shown the molecules that worked and those which did not, allowing them to improve their predictions in the future.