Loop fix?

Case number:699969-2011851
Topic:General
Opened by:alcor29
Status:Open
Type:Suggestion
Opened on:Friday, July 9, 2021 - 20:49
Last modified:Sunday, July 11, 2021 - 05:31

How about making the "penalty" for bad loops MUCH bigger so scripts don't mess up the good loops we started with?

(Fri, 07/09/2021 - 20:49  |  2 comments)


HuubR's picture
User offline. Last seen 1 hour 5 min ago. Offline
Joined: 04/13/2020
Groups: Gargleblasters

This is certainly an interesting idea!  I would expect, though, that the amount of the penalty was not chosen at random.

Would it help to have a setting like "ideal loop importance", that you could set to a value ("much") greater than 1?  (Without increasing the +500 points in the ideal situation, of course ;-)

Having said that, I don't think it is the script that messes up a good loop.  The way I look at it (disclaimer: without having a background in biochemistry!), it is hydrogen bonds and hydrophobic interaction that act on the helices or sheets adjacent to the loop, and sometimes pull them in the wrong direction.  A higher penalty might stop a recipe in its tracks, but it will not change the tendency of that section of the protein to move in the "wrong" direction.

jeff101's picture
User offline. Last seen 3 hours 49 min ago. Offline
Joined: 04/20/2012
Groups: Go Science
I think it would also help to have the Ideal Loops filter penalty 
depend on the # of segments that are in non-ideal loops. This way
one non-ideal loop containing 14 segments would have a larger
penalty than one non-ideal loop containing 10 segments or two 
non-ideal loops each containing 3 segments. Having the penalty 
depend on the # of segments that are in non-ideal loops would 
help recipes fix non-ideal loops in a more gradual and incremental 
manner. 

Please see the comments below for more details:
https://fold.it/portal/node/2005262#comment-41271
https://fold.it/portal/node/2005262#comment-41273
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