only one template working in puzzle 283?

Case number:845829-987562
Topic:Crash/Hang
Opened by:Perzik
Status:Closed
Type:Bug
Opened on:Thursday, April 8, 2010 - 09:04
Last modified:Wednesday, April 14, 2010 - 08:27

In puzzle 283 I only have the template with initial score 190 that works. The other ones result in one stretched bond after threading (see picture). That stretched loop does not resume to normal when wiggling. Also the score does not go above 0, even when wiggling it till there are no clashes etc anymore. Also, when local wiggling the loop, the stretched bond is going to normal size. But in local wiggle it behaves like the clashing importance is set to zero for that part. The other loops behave normally in local wiggle.
I tried all templates, retried the one that worked, tried closing and restarting foldit..
Anyone seen that problem?

AttachmentSize
283.PNG298.6 KB
(Thu, 04/08/2010 - 09:04  |  3 comments)


infjamc's picture
User offline. Last seen 4 hours 7 min ago. Offline
Joined: 02/20/2009
Groups: Contenders

Perzik:

The problem you're observing is happening because there is a huge gap between the target sequence and the template (6 for the 190 alignment and 8 for the remaining four). The reason that wiggling doesn't help is that atom-atom bonds are stretched to unreasonable lengths in the attempt to force the protein into the structure of the template (this is why the backbone score would be in the negative millions). To fix it, rebuild the 69-74 section, and the backbone score should drop to the negative hundreds at worst (although the clashing score could still be in the negative thousands... but at least that is fixable)

Foldit developers: problems like this is exactly why the ability to add and remove cut-points (which is already being used in certain existing design puzzles) could be useful (so that we can rather leave a gap in the protein and then sew the loop back together later rather than stretching bonds like this...)

Perzik's picture
User offline. Last seen 3 years 50 weeks ago. Offline
Joined: 05/13/2008
Groups: Boinc.be

Thanks for the explanation. I know why it happens (and it is indeed normal to first start with a better alignment), but I found it strange that the stretched bond was not 'destretched' after threading, that foldit would actually allow such stretching. Also, wiggling did not 'sense' that stretching, but local wiggle did..it was a strange behaviour that's all.

Good idea about the cutpoints. Another handy thing would be to use different templates on the same protein at the same time. If one template is too short and leaves a gap, and another template does fill that gap but misses other regions..you could use a part of one template and a part of the other..

beta_helix's picture
User offline. Last seen 6 hours 38 min ago. Offline
Joined: 05/09/2008
Groups: None

It was weird that Foldit would allow such stretching (when normally Foldit fixes that problem), but the problem in this case was that the solved template structures had that stretched bond as well!

I discuss this here: http://fold.it/portal/node/987585#comment-6824

I will bring up the cutpoints idea at our next Foldit meeting, thanks!

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, Boehringer Ingelheim, RosettaCommons