1902: Aflatoxin Challenge: Round 18 Loop Prediction

Closed since over 5 years ago

Summary

• Created: October 09, 2020
• Expires: October 16, 2020 at 18:00 UTC
• Max points: 100

Description

Refold the loops in this Aflatoxin Challenge design! In the most recent lab tests the Siegel Lab identified three well-behaved designs that show activity against aflatoxin. We want to know how these designs fold in the absence of aflatoxin. This puzzle features a protein designed by LociOiling in Puzzle 1739, but the starting structure has been partially unfolded. Help us predict the structure of this design by folding the redesigned loops!



Aflatoxins are a class of poisonous compounds that contaminate a significant portion of the global food supply. In the most recent phase of Foldit's Aflatoxin Challenge, the Siegel Lab at UC Davis proposed a new strategy for degrading aflatoxin, asking Foldit players to help redesign an oxidation enzyme so that it could break down aflatoxin molecules. In this puzzle, players can refold the two binding loops of the redesigned enzyme. The rest of the protein is frozen, except for some sidechains that might interact with the binding loops. No mutations or insertions are allowed. Since we are interested in how the protein might fold without aflatoxin, there is no aflatoxin molecule in the binding pocket.



This is the 18th round of our Aflatoxin Challenge, sponsored by Mars Inc. and Thermo Fisher Scientific. Promising designs will be tested by the Siegel Lab at UC Davis. By participating in the challenge/game, the players agree that all player designs will be available permanently in the public domain, and the players will not seek intellectual property protection over the designs created as part of the challenge/game.

Top groups

• 1. Beta Folders 100 pts. 9,932
• 2. Go Science 79 pts. 9,925
• 3. Gargleblasters 61 pts. 9,886
• 4. Anthropic Dreams 47 pts. 9,862
• 5. FoldIt@Netherlands 35 pts. 9,843
• 6. L'Alliance Francophone 26 pts. 9,832
• 7. Contenders 19 pts. 9,828
• 8. Russian team 14 pts. 9,735
• 9. Hold My Beer 10 pts. 9,731
• 10. Marvin's bunch 7 pts. 9,710

Top soloists

• 1. LociOiling 100 pts. 9,932
• 2. mirp 98 pts. 9,925
• 3. NeLikomSheet 95 pts. 9,915
• 4. Xartos 92 pts. 9,893
• 5. Mike Lewis 90 pts. 9,886
• 6. ichwilldiesennamen 87 pts. 9,881
• 7. ZeroLeak7 85 pts. 9,880
• 8. KarenCH 82 pts. 9,862
• 9. GuR0 80 pts. 9,853
• 10. Hiro Protagonist 78 pts. 9,843

Comments

[BootsMcGraw]

You say there are two binding loops, but only one is obvious to me. Can you please identify the positions of the two loops?

[malphis]

I think the second is 105-114. When I tug on it the score changes. It is also colored as opposed to the greyed locked area, but color blindness might interfere with seeing it. The other is 35-65.

[LociOiling]

Malphis is correct, the moveable parts are:

• 36-65
• 105-114 </ul> These are the moveable parts, with unlocked backbone. They can't be mutated, however, so they're not "designable". If you select EnzDes coloring and Cartoon or Cartoon Thin, there should be a contrast between these sections and the locked part. In color blind mode, the moveable parts show in in a lighter color with these settings. To me, they're a lighter green, while the locked parts are kind of gray-green. There are also several sections with locked backbone, but moveable sidechains. The easiest way to see these sections is to use the selection interface, the do a control-a to select all. The moveable backbone parts and the moveable sidechains parts then show up in blue, again a good contrast with the totally locked part. The recipe print protein also identifies these sections as segment ranges. The latest version uses Lua table format:

  unlocked_sidechain = {
      { 4, 14, },
      { 16, 16, },
      { 24, 25, },
      { 32, 32, },
      { 35, 71, },
      { 81, 81, },
      { 83, 83, },
      { 85, 98, },
      { 101, 103, },
      { 105, 114, },
      { 130, 133, },
      { 143, 143, },
      { 149, 149, },
      { 155, 155, },
      { 158, 158, },
      { 160, 161, },
      { 168, 168, },
  }
  

  The table includes the moveable backbone parts. When you show all sidechains, these unlocked sidechains should show as more lighter or more colorful, while backbone will be darker where it's locked. (Edit: corrected first moveable segment, it's 36, not 35, and also this is not a design puzzle, so it's "moveable", not "designable".)

[BootsMcGraw]

Thanks to Loci and malphis for the clarification.

Grey, green… they look the same to me. I'll just have to remember the numbers.

[LociOiling]

It looks like the first moveable section actually starts at 36, not 35. Also, 1902 does not have any mutable segments, so "designable" was not right. It's actually the "moveable" parts we're talking about, unlocked backbone or sidechains. I've made some corrections to the previous post.

Also, while I shared a snapshot from Puzzle 1739, the sequence for that one is not the same as the sequence for 1902. Segments 105-107 are "rds" in 1902, but "kdq" in my best from 1739. Segments 55 and 58 are also different 55 "q" in 1902, was "l" in 1739, while 58 is "a", was "l".

I shared several solutions with scientists on 1739, but so far I haven't found one with same sequence as 1902. I'll keep digging and share anything I find.

Finally, the print protein output shows an number of differences between 1739 and 1902 in which segments have locked backbone and sidechains. Also, the permanent cutpoints appear to be in different spots in the two puzzles, but print protein can't detect this.

[Formula350]

Press "F" to Freeze EVERYTHING.
If the backbone and sidechain becomes Frozen, then that is a FULLY "moveable" segment. (Can cut them, change their positions, and their structures)
If only the sidechain becomes Frozen, then all you can do is Shake or Pull on those sidechains because the backbone is Locked.


ALTERNATIVELY…
You can also do one other thing, but it requires Selection Interface: Toggle the "Move Tool" to displace and with nothing selected (press spacebar a couple times to make sure), on the Arrow-Cross you Right-Click and drag the mouse to move (or ctrl+left-click) which will move ALL of the segments we can manipulate.

It's crude, but since they are only connected by un-closable cuts, they're free to be relocated anywhere and will snap-back into place when wiggled.

[LociOiling]

I checked all my leftovers from puzzle 1739, and I still couldn't find an exact match for the amino acid sequence in 1902.

[jeff101]

Perhaps the sequence for 1902 actually did come from
one of your 1739 solutions, but it was from one of the
intermediate solutions shared automatically every few
minutes to Foldit instead of one of your Share with
Self, Share with Group, or Share with Scientists
solutions.

[jeff101]

https://fold.it/portal/node/2010518
says 2 LociOiling and 1 Phyx design
from Puzzle 1739 did well. Maybe
1902 is really a Phyx design from
1739.

[bkoep]

I don't know exactly how the Siegel Lab selected solutions, but they were looking at all of the solutions we collected from Puzzle 1739 (including intermediate solutions that are automatically uploaded during gameplay). This may not be a solution you saved in your client. Though I've verified that this solution was indeed created by LociOiling.