Sheet scoring on puzzle 969 doesn't work properly.

Case number:671071-998319
Topic:Game: Other
Opened by:BootsMcGraw
Opened on:Friday, August 15, 2014 - 02:23
Last modified:Monday, August 18, 2014 - 13:46

Something is hosed with the sheet bonus/penalty function on puzzle 969. The instructions say that designing a protein with greater than fifty percent sheets will result in a five hundred point bonus, while having less than fifty percent sheets will be penalized.

I have changed my design to have anywhere from zero to seventy-five segments set as sheets, and no matter how many segments I changed to sheet, I am penalized almost four thousand points.

I am sure this is not what you intended. Please fix this.

(Fri, 08/15/2014 - 02:23  |  4 comments)

Joined: 12/06/2008
Groups: Contenders

OK... with the help of some of the other players, I've figured it out.

A sheet is recognized by a hydrogen bond, not by setting the structure to "sheet". The scoring function considers a hydrogen bond to be equivalent to a sheet, even if the structure isn't set as sheet.

So, no bug. But I am not closing this feedback for awhile, since I was not the only person in chat who wa confused about the scoring.


bkoep's picture
User is online Online
Joined: 11/15/2012
Groups: Foldit Staff

Hi Boots, just to clarify: Have you changed the shape of your protein so that multiple strands align to form regular hydrogen bonds? Or have you simply selected residues and changed their sheet/helix/loop designations in the Foldit interface?

The Secondary Structure filter uses the DSSP method to identify hydrogen-bonding patterns typical of sheet and helix secondary structures. The filter is ignorant of the secondary structure designations that players can assign in the Foldit interface. I realize we've only mentioned the mechanics of the Secondary Structure filter in passing, but since it looks like the filter is here to stay we'll make a blog post explaining exactly how it works.

spmm's picture
User offline. Last seen 7 weeks 4 days ago. Offline
Joined: 08/05/2010
Groups: Void Crushers

Maybe an example of what you are looking for as well - I'm sure this is basic biochem and protein structure for those who have studied in the area but it is not that simple for those who haven't. Probably if you know where to look there are some great examples in the PDB but that would seem to miss the point.

Do hydrogen bonds from glycine, for example, to a 'sheet' count? 50% pure bonded sheets with core filter is hard work in 75 segs.

Also the puzzle is really stiff and very, very slow, stopping wiggle or shake is unresponsive for seconds, (on a good end point) ie noticeable and unresponsive - 'design' is not what springs to mind in this puzzle, it is impossible to move the backbone even on CI=0.

Joined: 09/21/2011
Groups: Void Crushers

See my share with the scientists. Plenty of connected sheets but filter says differently.


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