On 672 type puzzles

Case number:699969-994465
Opened by:Timo van der Laan
Opened on:Sunday, February 3, 2013 - 15:11
Last modified:Saturday, February 9, 2013 - 12:23

I started with a good symmetry. Found the docking places.
I really think that if this was posted not with a froozen part but with constaints on the mirror part (tight ones on the core, no so tight in the extremes) we could come very close to the native. And some bonussystem on conformance with symmetry is really needed in this type of puzzle.
The froozen part of this one was so bad, especially the sidechains, that it is not realistic in my opinion.
I think in the native, there will be 2 cystine bridges.
When I tried that I could not get the score above 6600.

(Sun, 02/03/2013 - 15:11  |  3 comments)

Susume's picture
User offline. Last seen 4 days 9 hours ago. Offline
Joined: 10/02/2011

This illustration of a protein with the same sequence http://en.wikipedia.org/wiki/File:PDB_2ezw_EBI.jpg shows at least 12 slightly different positions for the backbone, suggesting either that the real protein is somewhat flexible or that there is error variance in the measurement of it. Either way it makes sense to model the symmetry target as a movable protein with constraints rather than a frozen one. If there is some reason it must be presented as frozen, please at least shake and wiggle it with constraints before freezing. When it is presented with clashes and a stressed backbone, the foldit score function constantly tries to "fix" our copy by making it less symmetric.

marie_s's picture
User offline. Last seen 1 year 34 weeks ago. Offline
Joined: 05/18/2008
Groups: None

I disagree with more specific score function.
In my opinion, it will alter the learning process and the intuition.
If we have to work on docking, give the start with 2 symmetric proteins, (no need to a useless pseudo qttn) and let us play as usual.

Joined: 08/24/2011

While I agree completely with Marie's opinion, I must add that we probably won't have another 672. Wasn't that puzzle a simple workaround of the symmetry puzzles problem? As 675 behaves (mostly) well, I guess we can forget about this peculiar tinkering.


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