Place GDT numbers on puzzle's pages

Case number:845799-987516
Opened by:Madde
Opened on:Sunday, March 28, 2010 - 16:11
Last modified:Thursday, July 8, 2010 - 14:14

Place the GDT numbers (once they're available) next to the scores on the puzzle's pages.
Maybe in this way we could find some hidden folding talents. I could think of a pretty close to the native solution that nevertheless scores badly because the folder didn't have the patience or time to squeeze out the last 200 points out of the structure. (Sometimes a small change causes a huge difference in score.)

(Sun, 03/28/2010 - 16:11  |  7 comments)

Joined: 04/19/2008
Groups: None

Good suggestion. I haven't heard of GDT before but I am a bit familiar with RMSD (from looking at the screensaver of Rosetta@home). You could place the starting RMSD and the RMSD of the lowest energy model on the puzzle pages when they are closed. That way, we could see if and how much we have improved the protein.

Madde's picture
User offline. Last seen 1 year 5 weeks ago. Offline
Joined: 05/29/2008
Groups: Void Crushers

Since the natives of CASP9 puzzles 1 and 2 are there now, I want to bring this suggestion back to your attention. Could you please post the GDT at least of the 5 submitted solutions of each puzzle?

Joined: 09/18/2009
Groups: SETI.Germany

What is GDT?

Madde's picture
User offline. Last seen 1 year 5 weeks ago. Offline
Joined: 05/29/2008
Groups: Void Crushers

"[...] GDT (Global Distance Test), which is the metric used by the assessors in CASP (the higher the number the closer to the native) [...]"

beta_helix in

Joined: 05/09/2008
Groups: None
Status: Open » Closed

Great idea Madde. Where would you like these posted?
Do you want to just add them to your awesome scoreboard?

Here are the results from the first 2 puzzles:


MODEL1 from PUZZLE_1A, Mark-, Contenders: GDT=77.6 Template_used: 1fx4A
MODEL2 from PUZZLE_1B, aap, Richard Dawkins Foundation: GDT=70.2 Template_used: 1wc1A
MODEL3 from PUZZLE_1B, smith92clone, Anthropic Dreams: GDT=65.4 Template_used: 1wc1A
MODEL4 from PUZZLE_1A, Marie Suchard: GDT=75.6 Template_used: 1fx4A
MODEL5 from PUZZLE_1C, Madde, Void Crushers: GDT=72.3 Template_used: 1azsB


MODEL1 from PUZZLE_2D, aap, Richard Dawkins Foundation: GDT=61.4 Template_used: 3lyxA
MODEL2 from PUZZLE_2A, mimi, Contenders: GDT=70.8 Template_used: 2v1aA
MODEL3 from PUZZLE_2C, keypad5, Richard Dawkins Foundation: GDT=48.7 Template_used: 1bywA
MODEL4 from PUZZLE_2D, TheGUmmer, Void Crushers: GDT=78.2 Template_used: 3lyxA
MODEL5 from PUZZLE_2B, Mark-, Contenders: GDT=63.5 Template_used: 2pd7A

Obviously most of these models had more than one player contributing to it (and in the models we submitted to CASP we include every player that worked on that model in the REMARKS section) but here we just show who got the final top score.

Remember, the higher the GDT the better (100 would be a perfect match the native, 0 would be very bad!)

Madde's picture
User offline. Last seen 1 year 5 weeks ago. Offline
Joined: 05/29/2008
Groups: Void Crushers

Thank you very much!
I added them to the CASP9 Puzzle List:

Waiting for the next results :)

Joined: 09/18/2009
Groups: SETI.Germany

Good job, guys!

Thanks for the info.


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