Locked parts are not locked in latest puzzle

Case number:845813-989227
Topic:Game: Tools
Opened by:Sidran
Status:Open
Type:Bug
Opened on:Friday, February 4, 2011 - 04:42
Last modified:Thursday, October 27, 2011 - 05:13

Picture says it all. I noticed that sometimes "locked" parts move relative to each other.
I saw this 3-4 times on a single puzzle but only this time actual clashes appeared.

AttachmentSize
foldit_1296781363.png540.67 KB
(Fri, 02/04/2011 - 04:42  |  18 comments)


beta_helix's picture
User offline. Last seen 1 day 12 hours ago. Offline
Joined: 05/09/2008
Groups: None

What tools were you using when the "locked" parts moved?

Sidran's picture
User offline. Last seen 7 years 28 weeks ago. Offline
Joined: 12/15/2009
Groups: None

Yes, it is somehow connected to usage of bands.
But I wouldnt stop investigation there.
I cannot say it is only with bands. I am not looking
at locked parts when working.

Joined: 06/17/2010

Locked parts are moving when wiggle and there is some big clash.
Also strong bands can pull/push locked parts when wiggle.

spmm's picture
User offline. Last seen 48 weeks 2 hours ago. Offline
Joined: 08/05/2010
Groups: Void Crushers

They have alwasy moved for me :-) - since October 2010 I put bands on sheets to see the effect it has on the score

Joined: 05/03/2009
Groups: Contenders

I'm not sure of the issue here, it may not be a bug, but may be news to FoldCentral - greyed out 'locked' backbones will always move and shift with a band approach, when there are any predefined gaps in the grey backbone, even when constraints are imposed. The only way to lock movement in the grey backbone puzzles is to fix them in formaldehyde, and have a continuous structure, so it's like concrete. Free chains can then dance, if that's the wish of a puzzle.

tldr; constraints arent as powerful in fixing grey backbone puzzles as they want to be

CFC

Joined: 06/17/2010
Topic: Biochem » Game: Tools

Locked parts are moving when are really NOT do that.
In puzzle 399 cant use any LWS because entire thing is floating around.
Look at this screen:
http://fold.it/portal/files/chatimg/irc_174969_1297793584.png
Made short script to catch this.
Wiggling locally one segment (316) and testing distance on "other side" (311-312).
IT IS CHANGING during LWS even if 315 and 317 are frozen.
Also score drops rapidly because of constrains.
IMO this shound not have place in any puzzle...

Sidran's picture
User offline. Last seen 7 years 28 weeks ago. Offline
Joined: 12/15/2009
Groups: None

Frequently, its visually obvious that fixed parts move just a bit (observing them while doing operations pretty unrelated to them). Lately, I observed such moves without ever using bands.

This numeric example of yours could be another way to show that something deeper inside is not working as it should.

Its fascinating how long such a critical issue takes for resolution. Either we could be wrong and would be given explanation why (case closed) or whole case would not be still "open" hanging in the air.

beta_helix's picture
User offline. Last seen 1 day 12 hours ago. Offline
Joined: 05/09/2008
Groups: None
Status: Open » Closed

This should be fixed from now on, please let us know if it happens again.

Thanks for being so patient with this.

Sidran's picture
User offline. Last seen 7 years 28 weeks ago. Offline
Joined: 12/15/2009
Groups: None
Status: Closed » Open

This time its 402: Design the interface 8

Luckily, I know exactly what I done as it happened on a very beginning.

Started puzzle, wiggle, reset puzzle, added 3 backbones, wiggle, removed 1 backbone and left it on wiggle while I was watching some documentary. Screenshot is showing what I encountered when I switched back. Not just that "locked" parts moved a little and created clashes, but that locked part connecting locked chunks with modifiable one got strangely prolonged (as seen on ss).

From the very start there were no bands (outside those two imposed by puzzle), behavior changing, rebuilds, mutate or anything else (not even shake).

In attachment is exactly the same situation from another angle.

Sidran's picture
User offline. Last seen 7 years 28 weeks ago. Offline
Joined: 12/15/2009
Groups: None

Addition to previous report:
Its visible on ss but its worth mentioning that I added and deleted backbones on just one branch. The other I left untouched. I tried and I successfully replicated similar behavior.

Joined: 06/17/2010

Yap, same error. Local wiggle on sheet side causes huge score drop on loop side and entire locked fragment are moving.

Joined: 12/27/2010
Groups: None
Joined: 09/27/2011

My entire 471 puzzle moved and created several clashes inside the locked areas.

Seen here: http://fold.it/portal/files/chatimg/irc_332653_1319426583.png

beta_helix's picture
User offline. Last seen 1 day 12 hours ago. Offline
Joined: 05/09/2008
Groups: None

what were you running?
wiggle all, local wiggle, a particular recipe?

is it reproducible?
ie if you run a certain recipe then it will always cause the frozen area to move?
you can imagine that it is hard for us to reproduce this bug (since it rarely seems to happen) so any additional info you can give us would help!

Thanks!

Joined: 12/27/2010
Groups: None

On 471, if you do a rebuild the whole of the variable section such that it becomes strongly red, the whole protein sometimes moves as 2 separate pieces on wiggle. You can also try moving the ligand and doing the same. This protein often moves the locked parts so you should be able to reproduce it.

Joined: 12/27/2010
Groups: None

Another way to demonstrate the moving of locked parts on 471 - reset the puzzle, set clashing importance to 0 then wiggle. It will move the locked parts.

beta_helix's picture
User offline. Last seen 1 day 12 hours ago. Offline
Joined: 05/09/2008
Groups: None
Status: Open » Open

Thank you, that totally did the trick and it looks like we'll need to add a cutpoint to the opposite end of the loop in order to avoid this.
I apologize for not noticing this bug beforehand, but hopefully it won't happen in Loop Remodel Design Puzzle 2... hopefully...

Joined: 12/27/2010
Groups: None
Sitemap

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