Placeholder image of a protein
Icon representing a puzzle

1446: Y1 Receptor Ligand Docking

Closed since almost 5 years ago

Intermediate Overall

Summary


Created
October 31, 2017
Expires
Max points
100
Description

This is a follow-on puzzle of Puzzle 1441: Y1 Receptor Homology Modeling. In the previous puzzle, we asked you to model the structure of the human Y1 neuropeptide receptor. In this puzzle, we'd like you to determine how a known binder of Y1 interacts with that model.

We're starting you off with the top scoring structure from puzzle 1441, made by reefyrob and LociOiling.
We've placed BIBP-3226 into the homology model, in the approximate location of the binding pocket between the transmembrane helicies.
BIBP-3226 is a potent antagonist for Y1, and is speculated to bind in this general location.
For this puzzle, we'd like you to find the best location and best interactions of BIBP-3226 to Y1.

As proteins sometimes rearrange slightly around a small molecule, we've allowed for some flexibility in the protein,
but as we'd like for you to focus on how the ligand binds, you'll get a penalty if the protein moves too far from the starting model.

Top groups


  1. Avatar for Go Science 100 pts. 13,658
  2. Avatar for Anthropic Dreams 2. Anthropic Dreams 77 pts. 13,656
  3. Avatar for Contenders 3. Contenders 58 pts. 13,635
  4. Avatar for Beta Folders 4. Beta Folders 43 pts. 13,625
  5. Avatar for L'Alliance Francophone 5. L'Alliance Francophone 31 pts. 13,621
  6. Avatar for Deleted group 6. Deleted group pts. 13,612
  7. Avatar for Gargleblasters 7. Gargleblasters 15 pts. 13,605
  8. Avatar for Void Crushers 8. Void Crushers 11 pts. 13,597
  9. Avatar for Marvin's bunch 10. Marvin's bunch 5 pts. 13,574

  1. Avatar for Bruno Kestemont
    1. Bruno Kestemont Lv 1
    100 pts. 13,658
  2. Avatar for alcor29 2. alcor29 Lv 1 98 pts. 13,656
  3. Avatar for Galaxie 3. Galaxie Lv 1 95 pts. 13,643
  4. Avatar for Susume 4. Susume Lv 1 92 pts. 13,640
  5. Avatar for Bletchley Park 5. Bletchley Park Lv 1 89 pts. 13,635
  6. Avatar for ZeroLeak7 6. ZeroLeak7 Lv 1 87 pts. 13,634
  7. Avatar for georg137 7. georg137 Lv 1 84 pts. 13,627
  8. Avatar for LociOiling 8. LociOiling Lv 1 81 pts. 13,625
  9. Avatar for christioanchauvin 9. christioanchauvin Lv 1 79 pts. 13,621
  10. Avatar for benrh 10. benrh Lv 1 77 pts. 13,621

Comments


Susume Lv 1

Back when foldit used to give us Rosetta predictions to refine, the players complained the starting poses were too stiff in foldit because of being pre-optimized by Rosetta. The devs started giving us a less optimized version of the same shape (not sure if they perturbed them, or pulled them from an earlier point in Rosetta's run, or what). If the solutions to this puzzle show too little variation, you might try giving us an earlier save of reefyrob & Loci's with similar shape, which would be less stiff.

rmoretti Staff Lv 1

I realize that I could have been clearer in the description, but the primary goal of this puzzle is to move the ligand to find the best location and orientation of the ligand. We're less interested in protein rearrangements, and more interested in the position, orientation, and interactions of the ligand.

To clarify on the background, we know the structure of the ligand and that it binds the protein, but we really don't know exactly where it binds, or how it binds, or the internal conformation it takes when it binds. The ligand was arbitrarily placed in the protein with an arbitrary conformation, and we hope that players can find where it should go.

As the goal of the puzzle was primarily to find the position and orientation of the ligand, and not the internal structure of the protein, providing an optimized protein with structural restrictions was deliberate, so that players could focus on the ligand position, rather than having to re-optimize the protein structure. I did not completely freeze the protein for two reasons. The first is that predicted structures tend to be a little bit inaccurate, and adding a ligand might give additional information which can help correct some of the minor inaccuracies. The second is that proteins sometimes flex and move a little when ligands bind, so even if we had a perfect structure we would need to accommodate that.

Susume Lv 1

Thanks for the clarification!

Tips on varying the shape of the ligand:

The ligand in this puzzle (unlike other recent ligand puzzles) has "rotamers", which are set positions it can bend into, just like a sidechain has rotamers. With sidechains, the number of possible rotamers changes when the backbone shape changes. With the ligand, I believe the number of possible rotamers will remain constant because it does not have a backbone to change - someone could test this and post results if they want. My ligand and LociOiling's ligand both report 49 rotamers (positions).

The Shake function switches sidechains and the ligand from one rotamer to another if it can find a combination it thinks is better. I believe shake tries to obey rubber bands on the sidechain or ligand if they are present, so you could try banding and then shaking if you have a particular shape in mind. Edit to add: Wiggle All can also bend the ligand, though not as much as switching rotamers can.

To manually switch rotamers, use selection interface, select (highlight) the ligand by clicking on it, and use the left and right arrow keys to step through the rotamers.

Some scripts explicitly cycle sidechains through their rotamers. Those scripts can also change the ligand position IF they operate on the ligand - many foldit scripts are written to ignore ligands, so have a scripter check the one you are thinking of using. Players can help out by posting here names of public scripts that do operate on the ligand and cycle through rotamers.

The lua functions rotamer.GetCount(segnum) and rotamer.SetRotamer(segnum, rotnum) can be used to check how many rotamers there are and to set the ligand (segnum 320) to a particular rotamer. I do not know if a particular rotamer always has the same number - it depends whether the game sorts them somehow as it pulls them out of the database.

LociOiling Lv 1

I created a simple script to flip through all the rotamers for segment 320. So far, the count has remained the same. I'm not sure if the ligand's position relative to the protein could affect the rotamer count, however.

LociOiling Lv 1

Thanks to jeff101 for figuring out that the ligand in this puzzle is BIBP-3226, which was designed for the Y1 receptor.

See wikipedia: https://en.wikipedia.org/wiki/BIBP-3226

In JMol, BIBP-3226 looks like this:

[img_assist nid=2004438 title=BIBP-3226 seen in JMol desc= link=node align=left width=640 height=457]

In Foldit, it's difficult to get the ligand into exactly the same shape as seen in JMol. Here's a reasonably good effort:



[img_assist|nid=2004439|title=Puzzle 1446 ligand.|desc=The ligand in puzzle 1446 is the designer molecule BIBP-3226.|link=node|align=left|width=640|height=514]

LociOiling Lv 1

For some reason, Foldit shows 67 atoms for ligand segment 320. The formula I get is C27H32N5O3, while BIBP-3226 is C27H31N5O3, at least according to wikipedia.

Looking at the images above, it's not to hard to spot the difference. At the top, there are three blue nitrogens with a gray carbon in the center. JMol has double bond between the first nitrogen and rest of the molecule. Foldit has a single bond and an extra hydrogen instead. Foldit has the double bond is on the next nitrogen, counter-clockwise in this view.

In technical terms, it appears that the Foldit version may be a little goofed-up.

bkoep Staff Lv 1

There is indeed an "extra" hydrogen in the Foldit ligand, but this is not an error.

In water, at physiological pH (close to neutral; pH 7), this ligand will actually pick up an extra hydrogen. The guanidino chemical group (the carbon bonded to three nitrogens in plane) prefers this configuration, even though it means the molecule will have a net charge of +1. You might recognize the guanidino group also from the amino acid arginine (ARG), which behaves similarly—it also has an extra H and a net charge of +1 at physiological pH.

This behavior of the guanidino group is related to its double bond, which is a little bit strange in that it has several resonance forms. The electrons of the double bond are actually shared across all three C-N bonds of the guanidino group. Rather than thinking about the double bond as being between just one C-N pair, as depicted, it's probably better to think about the double bond as being "spread out" across all three C-N bonds. Likewise, the +1 charge is also spread out across the three nitrogen atoms; this is one of the reasons the configuration is so stable.

LociOiling Lv 1

The new recipe Atom Explorer can help to identify the atoms of a ligand or just a regular segment.

All it's really doing is drawing a spaceband to each atom, and recording your input as to what the atom is. Unfortunately, there's no way to tell which atom is which element with just a recipe alone.

The recipe outputs the results of your analysis in table format, which can be imported into another recipe.

For 1446, the table looks like this:

--
--    AtomChart indexes
--    ATCSEG = 1     -- segment
--    ATCATM = 2     -- atom
--    ATCELE = 3     -- element (user pick)
--    ATCBND = 4     -- bonding (user pick, D = donor, A = acceptor, B = both)
--    ATCCOM = 5     -- comment (user input)
--
atomChart = {
    { 320, 1, "C", "", "", },
    { 320, 2, "N", "D", "", },
    { 320, 3, "C", "", "", },
    { 320, 4, "C", "", "", },
    { 320, 5, "C", "", "", },
    { 320, 6, "C", "", "", },
    { 320, 7, "C", "", "", },
    { 320, 8, "C", "", "", },
    { 320, 9, "C", "", "", },
    { 320, 10, "C", "", "", },
    { 320, 11, "C", "", "", },
    { 320, 12, "C", "", "", },
    { 320, 13, "C", "", "", },
    { 320, 14, "C", "", "", },
    { 320, 15, "C", "", "", },
    { 320, 16, "C", "", "", },
    { 320, 17, "O", "A", "", },
    { 320, 18, "C", "", "", },
    { 320, 19, "C", "", "", },
    { 320, 20, "C", "", "", },
    { 320, 21, "N", "D", "OK, not goofy, but BIBP-3226 has double bond to next atom here", },
    { 320, 22, "C", "", "", },
    { 320, 23, "N", "D", "", },
    { 320, 24, "N", "D", "", },
    { 320, 25, "C", "", "", },
    { 320, 26, "O", "A", "", },
    { 320, 27, "N", "D", "", },
    { 320, 28, "C", "", "", },
    { 320, 29, "C", "", "", },
    { 320, 30, "C", "", "", },
    { 320, 31, "C", "", "", },
    { 320, 32, "C", "", "", },
    { 320, 33, "C", "", "", },
    { 320, 34, "C", "", "", },
    { 320, 35, "O", "B", "", },
    { 320, 36, "H", "", "", },
    { 320, 37, "H", "", "", },
    { 320, 38, "H", "", "", },
    { 320, 39, "H", "", "", },
    { 320, 40, "H", "", "", },
    { 320, 41, "H", "", "", },
    { 320, 42, "H", "", "", },
    { 320, 43, "H", "", "", },
    { 320, 44, "H", "", "", },
    { 320, 45, "H", "", "", },
    { 320, 46, "H", "", "", },
    { 320, 47, "H", "", "", },
    { 320, 48, "H", "", "", },
    { 320, 49, "H", "", "", },
    { 320, 50, "H", "", "", },
    { 320, 51, "H", "", "", },
    { 320, 52, "H", "", "", },
    { 320, 53, "H", "", "", },
    { 320, 54, "H", "", "", },
    { 320, 55, "H", "", "bonus hydrogen!!!", },
    { 320, 56, "H", "", "", },
    { 320, 57, "H", "", "", },
    { 320, 58, "H", "", "", },
    { 320, 59, "H", "", "", },
    { 320, 60, "H", "", "", },
    { 320, 61, "H", "", "", },
    { 320, 62, "H", "", "", },
    { 320, 63, "H", "", "", },
    { 320, 64, "H", "", "", },
    { 320, 65, "H", "", "", },
    { 320, 66, "H", "", "", },
    { 320, 67, "H", "", "", },
}

[edit: OK, Foldit view is not goofy, it just looks at those resonances differently]
[edit: missed atom 24, nitrogen donor]

brow42 Lv 1

I've created a GUI for banding to atoms: https://fold.it/portal/recipe/102533 . You can use this to help make hbonds, or deform the ligand. It works around built-in restrictions in the regular banding function. You can easily take out the function to use in other scripts, to remake your favorite bands, or random bands.