2 replies [Last post]
axcho's picture
User offline. Last seen 9 years 37 weeks ago. Offline
Joined: 12/18/2007
Groups: None

A few answers to some Foldit questions asked by jas0501. Thanks to beta_helix for getting this started, and to zoran for further details.

Has anyone ever attempted to configure a protein into the "known best" configuration using Foldit?
We've tried matching the "known best" configuration by using the native as a guide. As we suspected the difficulty depends on the secondary structure. Very often the sheets and helices can be matched perfectly, while it is harder to perfectly match loops since they have more wiggle room. In many cases we have been able to get the sidechains to line up perfectly with the native.

For the Grand Challenge puzzles:
1. Are any of the scores better than the known solutions?
Yes, there are a number very encouraging results where we have found the best solutions. For some more complicated cases, it seems that different soloists and groups have gotten different aspects right, but not necessarily the entire protein. This is why the All Hands challenges are crucial, as we expect that the in this stage all the best aspects from different solutions will be integrated together.

We're trying not to reveal too much about the Grand Challenge results until we've run through all the puzzles "because it could taint results for future puzzles that we post," but if you have any more specific questions we'll see if we can answer those safely. ;)

2. If not how close are the Folders coming to the optimal configuration?
As mentioned above, when we look at each specific part, we find that at least one person has folded it perfectly. We are hoping that a new feature we plan to introduce to All Hands mode will help combine the strengths of each solution more easily.

3. Have any of the solutions provide a novel configuration?
Yes, if you consider any case that isn't close to the native as a novel configuration. But then again, it's easy to create a novel configuration, while it's hard to find a very high-scoring novel configuration.

On a related note, the next big change we'll be rolling out after the Grand Challenge puzzles is a new set of tools and visualizations for protein design. The puzzles we release for protein design will be all about finding novel solutions. We are hoping that this will be an area where humans will really excel when compared to automated approaches.

Will there ever be a puzzle of a small protein in a straightened initial configuration?
We did this once last summer, and we can try it again if people are interested. We will introduce more of those after the complete All Hands for Grand Challenges. In general, computers do not have trouble with the initial stage of folding, at least for small proteins, so we have not attempted to explore that part of the folding space.

Are there any published papers outlining any observations and learnings gleaned from the study of the Foldit soltions?
We are currently working very hard on getting this done! This is what we're doing with the Grand Challenge puzzles. For more information, see these three posts on the Foldit blog. Nature has a press embargo on all submitted findings, so we can't put too many things out until they decide on the paper, but we will provide partial updates on the blog as soon as we synthesize all results.

Are there currently any PhD candidates that are basing their thesis around Foldit?
Yes. Seth Cooper is working on Foldit for his Computer Science Ph.D.

Has any consideration been given to patenting the Foldit UI approach and use of the internet community as a mechanism for solving protein folding problems?
We are interested in democratizing science, empowering as many people to actively take part in scientific research. We will do everything to enable this goal.  At this point it is not clear why a patent would help us achieve this goal. In fact, it could very likely be counterproductive.

Joined: 08/18/2009
Groups: None
Answers to General Foldit Questions

OOps- I gave this post a -1 before I realized it. Ignore this. It is a great post. :D

Joined: 05/09/2008
Groups: None
Thanks for the info! I have

Thanks for the info!

I have introduced Fold It to several friends, and the main question is "What the heck am I working on???"

Please do post some "small print" descriptions of the proteins and what they do for the puzzles. It makes Fold It that much more interesting. Don't worry too much about making the text "understandable by the masses". Any info will do, it makes working on the puzzles that much for meaningful to us. But if for some reason it interferes with the results (someone googleing the answer?) I can understand that.

Thank you for your work on this project!


Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Amazon, Microsoft, Adobe, Boehringer Ingelheim, RosettaCommons