The players were able to create what the scientists described as "models of sufficient quality" for the scientists to complete the structure.
The scientists hail from the University of Washington, A. Mickiewicz University in Poland, the Polish Academy of Sciences, and the Academy of Sciences of the Czech Republic. The crowdsourcing effort is described in a research paper, "Crystal Structure of a Monomeric Retroviral Protease Solved By Protein Folding Game Players," published in the current issue of Nature.
Five Out of Ten
The problem in defining this particular molecular structure has puzzled scientists for years. Protein folding is important in a variety of diseases, including Alzheimer's, cystic fibrosis, and mad cow disease. Mis-folded proteins have been related to disease, while properly folded ones could, for instance, block the HIV virus.
Foldit is an online collaborative game developed by researchers as a way to harness distributed computing and crowdsourced resources. Players win or lose points by creating viable structures within the established rules. Red balls with spikes, for instance, appear when sections of a protein are too close together, resulting in a loss of points.
In their Nature paper, Seth Cooper of the University of Washington's Center for Game Science, who was the developer of Foldit, said he and his colleagues knew that "top-ranked Foldit players excel at solving challenging structure refinement problems" relating to protein folding, even though the players are not scientists.
Ten puzzles involving protein-folding were introduced into the game, along with a scoring mechanism that related to how the proteins and their components should work. A player could increase a score by successfully re-twisting a protein, for example.
The scientists compared the Foldit results to their protein-folding prediction application, called Rosetta. The results: The collaborative players beat Rosetta on five of 10 protein-folding challenges. On three of the remaining, both efforts led to similar results, and on the other two, Rosetta was technically better but still "basically incorrect," the scientists said.
Collaboration and Competition
One of the most interesting results was how Foldit's process supported both collaboration and competition, resulting in both greater group and individual efforts. The researchers noted that players shared structures and exchanged solutions with their fellow group members, and helped each other out with strategies and tips via chatting and a wiki. Between groups, there was competitive motivation to play more.
Al Hilwa, program director for application development at IDC, called the Foldit effort and similar ones a "tremendous way to harness the power of the crowd, especially gamers, who are perceived to be doing nothing useful" as they spend countless hours solving game-based problems.
Hilwa said that, "once the software and models mature," he thought it "was absolutely possible" that smaller companies could begin using distributed processing and crowdsourced efforts on some commercial basis that would give them access to massive computing without the huge capitalization that is currently required.
Foldit is only one of several collaborative games intended to tap the crowd-based resources of "citizen science." Others include Galaxy Zoo, which uses more than a quarter-million Net-based users to classify hundreds of thousands of galaxies in the Sloan Digital Sky Survey; MoonZoo, designed for users ages 9 to 18 to design and build a virtual lunar rover; and SETI@Home, which uses distributed processing to look for intelligent signals from space.
No comments:
Post a Comment