Technical Highlight - November 2011
Short description: Foldit players successfully solve the crystal structure of a retroviral protease that had resisted previous determination.
According to the proverb, “All work and no play make Jack a dull boy,” but what if Jack could work and play at the same time? Recently, researchers gave players of the online game Foldit two separate opportunities to do just that. Foldit challenges players worldwide (no advanced understanding of protein structure needed) to solve protein structure puzzles, creating a platform that is amenable to crowd-sourcing protein modeling problems.
The first challenge was to use Foldit for challenges in the 2010 Critical Assessment of Techniques for Protein Structure Prediction (CASP). Foldit was entered as an independent group to see how players compared to other approaches. The results showed that Foldit players tended to focus on local energy minima, making them not as successful in most areas. However, when attempting refinement or de novo structural prediction, the Foldit Void Crushers group produced the best models in each of those categories in the CASP challenge, demonstrating the utility of the approach and mindset of Foldit players for these areas.
The second challenge was to use Foldit to predict the structure of a protein that has resisted structural determination. Over a span of 10 years, automated methods had repeatedly failed at solving the structure of the protease from the simian AIDS–causing Mason-Pfizer Monkey Virus (M-PMV PR). Baker and colleagues gave Foldit players 3 weeks to work with models derived from the NMR coordinates of M-PMV PR, and made sure that the starting models had poor Rosetta scores so as to avoid players being trapped in local energy minima. By the end of the challenge, one team, the Foldit Contenders group, amazingly produced a model that was successfully used to solve the crystal structure of M-PMV PR by molecular replacement.
The results of these two challenges highlight the potential of approaching structural problems differently, and also reveal pitfalls to be avoided. More importantly, the result of the M-PMV PR in particular demonstrates that online games and crowd-sourcing have the potential to provide fresh insights and solutions to long-standing problems.
F. Khatib et al. Crystal structure of a monomeric retroviral protease solved by protein folding game players.
Nat. Struct. Mol. Biol. (18 September 2011). doi:10.1038/nsmb.2119