720 wiggle sticks_stops

Case number:845813-995234
Topic:Game: Tools
Opened by:jamiexq
Status:Open
Type:Bug
Opened on:Tuesday, May 28, 2013 - 22:49
Last modified:Friday, June 7, 2013 - 02:28

WiggleAll on 720 sticks badly or stops altogether at CI of 1.00

(Tue, 05/28/2013 - 22:49  |  15 comments)


beta_helix's picture
User offline. Last seen 1 day 16 hours ago. Offline
Joined: 05/09/2008
Groups: None
Status: Open » Open

What OS are you using and is this when using WiggleAll by itself or within a script?

This doesn't happen from the starting conformation (that has many clashes), right?

We posted similar puzzles in devprev over the past 4 days and this was not an issue:
http://fold.it/portal/node/992823#comment-23183

jamiexq's picture
User offline. Last seen 18 hours 15 min ago. Offline
Joined: 07/31/2011

Thought I replied but do not see it - so here is my reply again
Using Win 7
wiggle sticks by itself, but also in scripts I think
wiggle at 0.91 got a 250 pt jump for me after the score bounce around for a while

But, all this said, no worries

I tend to do very well on broken puzzles

Thanks,

bkoep's picture
User offline. Last seen 18 hours 22 min ago. Offline
Joined: 11/15/2012
Groups: None
Status: Open » Open

I see what you mean about WiggleAll being "sticky", especially at the start when everything is isoleucine. Unfortunately, this may not be so much a programming bug as it is a symptom of the Wiggle algorithm.

As you might guess, Wiggle wasn't exactly designed to wiggle a glob of isoleucine (so maybe we shouldn't have used this as the starting structure for this puzzle...). It might be getting hung up trying to find a way to improve the score of something that should score very, very badly.

This is still a little strange, though, I'll keep looking. Does wiggle work any better for you after some mutating?

spvincent's picture
User offline. Last seen 18 hours 59 min ago. Offline
Joined: 12/07/2007
Groups: Contenders

I see the same 'stickiness' on a reasonably mutated solution (Max OS X): makes it very hard to play. Scientist-shared solution sv12260.

Joined: 04/19/2009

720 is truly broken.

I can't nail down all of it yet, but I tried to do what jamie said about 721 wiggle all at zero on this one.

My score was 9030 when I started the wiggle. The protein did crumple - but not before giving me a gain at the beginning of the wiggle. I did it twice more, for a net gain of 10 points and a rank increase from rank 10 to rank 5, score of 9040.

A zero point zero wiggle should never give a score increase!

bkoep's picture
User offline. Last seen 18 hours 22 min ago. Offline
Joined: 11/15/2012
Groups: None

Yeah, there's definitely some weird wiggle behavior going on here.

Auntdeen, can you "share with scientists" your solution that sometimes gains points with wiggle at 0.0? This sounds like it could have something to do with the filters...

The "sticky" wiggle is a different problem, I think, and might have to do with the puzzle symmetry. Is anyone experiencing this "sticky" wiggle on non-symmetric puzzles (718, 719, or 721)?

bkoep's picture
User offline. Last seen 18 hours 22 min ago. Offline
Joined: 11/15/2012
Groups: None

jamiexq, I saw your feedback about puzzle 721 freezing with wiggle at 0.0 (http://fold.it/portal/node/995243). It's behaving a little differently when I run it...

In puzzle 721, either wiggle doesn't start or it works pretty smoothly. In puzzle 720, wiggle starts ok but gets hung up pretty quickly and freezes consistently.

Are you seeing the same thing?

Joined: 04/19/2009

I've "shared with scientists" - but sorry, it's not at the stage when the zero wiggle happened. I got distracted and moved too far past the graph point.

I'd be happy to send you the log.txt also if that would help (I have 2 clients open atm so the log.txt will be a lot to look through).

beta_helix's picture
User offline. Last seen 1 day 16 hours ago. Offline
Joined: 05/09/2008
Groups: None

Did anyone who played the devprev puzzles (that have the same score function/filters/etc as this one) notice these same problems?

http://fold.it/portal/node/992823#comment-23183

either of these two puzzles:
http://fold.it/portal/node/995206/show_players
http://fold.it/portal/node/995207/show_players

Hopefully we'll be able to pinpoint where the major issue is, thanks for your patience!

spvincent's picture
User offline. Last seen 18 hours 59 min ago. Offline
Joined: 12/07/2007
Groups: Contenders

I looked at my old puzzle 708 and wiggle appeared to be working just fine.

beta_helix's picture
User offline. Last seen 1 day 16 hours ago. Offline
Joined: 05/09/2008
Groups: None

Thanks spvincent, this helps us narrow down when the issue occurred.

Can everyone who has run into this problem please try out this devprev puzzle and let us know if it happens there as well:

http://fold.it/portal/node/995207

I have re-opened both of these that we had posted after the update to main.

It would be very interesting if the issue doesn't occur on those puzzles!
Thank you all again for your help diagnosing this!

Susume's picture
User offline. Last seen 3 days 20 hours ago. Offline
Joined: 10/02/2011

Wiggle hangs for me on puzzle "BETA 76 Residue Dimer Design Test B (708)" - see share called "wiggle hangs", shared with scientists. Wiggle moves a few points and then stops, regardless whether I wiggle at full CI, CI 0.5, or CI 0.2. Details in the share comments. Win 7 64, latest devprev.

Susume's picture
User offline. Last seen 3 days 20 hours ago. Offline
Joined: 10/02/2011

Same result on "BETA 76 Residue Dimer Design Test A (708)" - see share called "wiggle hangs", shared with scientists.

beta_helix's picture
User offline. Last seen 1 day 16 hours ago. Offline
Joined: 05/09/2008
Groups: None
Status: Open » Open

Thank you all very much for sharing your sticky-wiggle solutions with us.

We've finally pinpointed the cause of the bug, thanks to your examples, and our priority is to resolve this nasty bug as soon as possible.

Thanks again for all your help and your patience with this one... we realize how annoying this bug is!

beta_helix's picture
User offline. Last seen 1 day 16 hours ago. Offline
Joined: 05/09/2008
Groups: None

Thank you to all of you for helping us with this nasty bug.

Please try out this devprev puzzle and tell us if the same issues occur:
http://fold.it/portal/node/992823#comment-23390

We really appreciate it!

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