5 replies [Last post]
Joined: 10/23/2014
Groups: Contenders

View the attached images for context.

I moved the top sheet down a little. The ideal loop (frozen) was no longer ideal. Using Auto Structures the top sheet gained another sheet segment. The ideal loop and attached sheets were not moved at all, just the top sheet above.

The question is when is an ideal loop not ideal loop? Is moving other segments supposed to turn ideal loops into non-ideal loops?

The frozen ideal loop was initially found using Remix though I think the size "randomly" increased from 3 to 4 segments later on.

Solution1.png312.51 KB
Solution2.png402.05 KB
Susume's picture
User offline. Last seen 1 day 45 min ago. Offline
Joined: 10/02/2011
filter should be more flexible

The filter was apparently happy with the loop when all four residues were classified (by the game) as loop. When one of them switched to sheet (not because its shape changed but because the sheet on the other side bonded to it) the filter rejected it. Since the shapes of the residues did not change, I think the filter should still give credit for the 4-residue loop.

Joined: 09/24/2012
Groups: Go Science
Excellent question

I think there must be something in the "auto-structure" that recognizes sheets and helices depending on bonds (and I suppose the bond quality or points, because you had the same number of bonds here).

If it changes above a certain threshold, it's possible that a frozen segment be recognized as (un)ideal loop.

Interesting to see that we don't need to close the cut in order to change this. A new way to try to idealize loops using cuts :)

Joined: 06/06/2013
Groups: Gargleblasters
Crazy things happening with score

I had an "ideal" loop with a backbone problem. Gitwut had an "ideal" loop with a low scoring kink and odd shape. We shared them on vet chat. This is not ready for prime time. I recommend more fix before it goes into something that scores

Joined: 09/24/2012
Groups: Go Science
Other "crazy" think

A suggested loop is not alway "rama" correct. I can understand that the system proposes us "AA neutral" shapes and that we have to mutate to the right AA afterward. But sometimes, it proposes an angle that just don't exist for any AA.

Susume's picture
User offline. Last seen 1 day 45 min ago. Offline
Joined: 10/02/2011
Maybe "ideal" is not the right word

I wish these loops had been named something else - not just because the word Ideal is already overloaded in foldit, but because it sets up certain expectations in the players that lead to frustration. Maybe these could be called Preferred Loops, or Approved Loops, or Rama loops, or Baker Loops, or Template Loops, or something else that doesn't make us think they will be perfect. They are just statistically better rama shapes - they might have other bad attributes that make them unlikely to fold right, or that make them have bad scores (such as backbone issues, poor ideality, clashes with neighboring parts of the protein), or that make them difficult to hold onto (such as having a non-glycine AA where only a glycine can be, or having fragile hydrogen bonds). Beating the filter means the loop matches a pattern of rama shapes on a short list of shapes the devs want to encourage - it doesn't mean the loop we have created in foldit is ideal, or stable, or high-scoring.


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