Three-dimensional structures of proteins are determined by their amino acid sequences, but protein structure prediction remains a complex problem that requires massive amounts of computing power. For years, people from all over the world have been donating their computers’ free processing time to Rosetta@home, which provides a screensaver that uses spare processor time to predict protein folding patterns. A group of researchers, however, thought that human intelligence might be better at solving these complex problems, so they developed a multiplayer online game, Foldit, in which accurate protein structure models are produced through gameplay. The game has an extensive motivation and reward structure with a score system, player statuses, ranks and forums. It turned out that humans were often very good at solving problems where computer algorithms would become stuck. Conversely, the computer performed better when starting from a simple linear chain of amino acids. Different players have different strengths, and by having team competitions, those strengths were combined to get the best results. For the article published in Nature, more than 57,000 players contributed through gameplay and feedback. The authors foresee this strategy being used in other scientific domains as well where human three-dimensional structural problem solving can be used. There is a video after the break.
More from Ars Technica: Gamers beat algorithms at finding protein structures…
Article abstract: Predicting protein structures with a multiplayer online game…
Project website: Foldit: Solve Puzzles for Science…