More Exciting Electron Density Results!

We wanted to update you on the Electron Density results for puzzles 746 & 752, now that the native solved structure was released for this CASP ROLL target.

Because we had also posted this puzzle previously for CASP ROLL target R0034, we can compare your results before we provided you with Electron Density information and after.

The first Foldit puzzle that we posted for this protein was Puzzle 717: CASP ROLL Target R0034, where we gave you the Rosetta Server predictions as starting models (generated using Rosetta@home).
On the RMSD plot below, you can see these starting models as black dots (and the black lines highlight how far they were from the native). The green dots represent your many different Foldit predictions, and for all these RMSD plots, you want to be as close to the left as possible (an RMSD=0.0 would be a perfect match to the native).
You can see that although there were many predictions that were closer to the native than the 5 starting models, these were NOT the lowest energy predictions (highest Foldit score).

The second Foldit puzzle that we posted for this protein was Puzzle 723: Server models for R0034, where we gave you 5 Zhang Server predictions as starting models. On the RMSD plot below, you can see these starting models as black lines (because the starting scores were so bad they don't show up on these plots). Again, there were many predictions that were closer to the native than the 5 starting models, but the closest ones were not the ones with the highest Foldit score (although the top-scoring solution was a bit closer to the native than the best Zhang Server prediction).

Then we gave you Puzzle 746: R0034 with Electron Density: Round 1, and suddenly the RMSD plot looks very different:

The first thing you might notice is how there are a lot fewer green dots compared to the 2 previous puzzles, and that is because most of the predictions for this puzzle are not on this plot (they all have an RMSD >10 ). But obviously the most exciting aspect of this plot, is how close some of you were able to get!

Starting from an extended chain (with untrimmed density that was too poor for the automated methods to solve this structure with) the top-scoring Foldit solution by players Xartos, svenmh, Schleicher, and Batz was amazingly close to the native!

Using their top-scoring solutions from Puzzle 746, we trimmed the density and reposted it as Puzzle 750: R0034 with Electron Density: Round 2.

Here are the final amazing results from that puzzle:

Below is a superposition of the native (in blue) with the top-scoring Foldit solution (in green) from Puzzle 750: R0034 with Electron Density: Round 2 by players Galaxie and Susume:

We are very excited about what you have been able to accomplish on such a difficult target, with no homologs or templates to help you!

We also hope the Black Belt Folding on ED was useful, as these RMSD plots have shown that ED is not easy for everyone!

We look forward to what you can do with density in the new Puzzle 756: Unsolved Salmonella Bacteriophage: CryoEM map, and in future unsolved cases. We hope the recent tools that have been added to the Electron Density panel have been useful, and we hope to improve and implement more of your suggestions soon!

Personally, I am very excited about the potential you have shown for solving proteins structures using experimental data that isn't good enough for automated methods to work with.
As I mentioned in the recent Scientist Chat I will be starting a faculty position in the CIS Department at the University of Massachusetts Dartmouth this fall, and this will be a focus of my lab: extending the Foldit framework beyond structure prediction by making it part of the experimental process for solving protein structures!
As the above results have shown, when we are able to give you experimental data, it makes reaching the native a lot easier for you!

This will be my last post this month, until I get settled in early September, so please do not be offended if I do not reply to your PMs. bkoep will be the lead scientist working on Foldit in the Baker Lab (so feel free to PM him and katfish while I am away!)

And as always: keep up the great folding!


( Posted by  beta_helix 81 1638  |  Thu, 08/08/2013 - 02:54  |  4 comments )
bkoep's picture
User offline. Last seen 4 days 7 hours ago. Offline
Joined: 11/15/2012
Groups: None

A shout out also to aendgraend, svenmh, Schleicher, and Batz, who produced a solution with an astounding 0.302 Å RMSD to the native structure. Unfortunately, our Foldit score function was not able to distinguish their solution from other top-scoring solutions. Still, this is a very impressive result.

steveB's picture
User offline. Last seen 2 years 31 weeks ago. Offline
Joined: 01/08/2009
Groups: Void Crushers
Upload FROM scientist feature

Quite an amazing sequence of results !

Could I make a request ?

When the natives for previous foldit de-novo puzzles are released, could you upload them to the server so that we can open up old puzzles, load in the native and compare it to our folds, and more importantly, to the ED cloud. That way, we can learn and improve !

Joined: 04/15/2012
Groups: Beta Folders

An amazing job all. Congrats!

Susume's picture
User offline. Last seen 1 day 7 hours ago. Offline
Joined: 10/02/2011
A team effort

A big thank you to karstenw, whose shared ideas were important in developing this solution. It was truly a team effort!

Get Started: Download
  Windows    OSX    Linux  
(10.7 or later)

Are you new to Foldit? Click here.

Are you a student? Click here.

Are you an educator? Click here.
Only search
Recommend Foldit
User login
Top New Users

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, RosettaCommons