Placeholder image of a protein
Icon representing a puzzle

1144: How low can you go? 10-residue Marburg binder design

Closed since almost 7 years ago

Intermediate Overall Design


October 01, 2015
Max points

This is the next in our series of Ebola/Marburg puzzles. We want you to design an ultra-small cyclic peptide to bind to and neutralize the Marburg virus glycoprotein. The glycoprotein is involved in viral entry into cells, and binding to a particular spot in the "head" region of this protein inhibits its interaction with human cells and abolishes its activity, neutralizing the virus.

We're giving you a ten-residue peptide with a disulfide bond linking the ends of the peptide. Two residues in the middle are taken from a loop of an antibody that binds to this head region. We want you to design the rest of the peptide to maximize favourable interactions, both within the peptide and with the Marburg glycoprotein. Pay particular attention to backbone hydrogen bonding -- the backbone hydrogen bond score has been upweighted for this puzzle.

Top groups

  1. Avatar for Contenders 100 pts. 12,844
  2. Avatar for Gargleblasters 2. Gargleblasters 83 pts. 12,843
  3. Avatar for Go Science 3. Go Science 69 pts. 12,842
  4. Avatar for L'Alliance Francophone 4. L'Alliance Francophone 56 pts. 12,839
  5. Avatar for Anthropic Dreams 5. Anthropic Dreams 45 pts. 12,808
  6. Avatar for Void Crushers 6. Void Crushers 36 pts. 12,808
  7. Avatar for Beta Folders 7. Beta Folders 29 pts. 12,800
  8. Avatar for BOINC@Poland 8. BOINC@Poland 23 pts. 12,791
  9. Avatar for Hun-Magyar Csapat 9. Hun-Magyar Csapat 18 pts. 12,789
  10. Avatar for FoldIt@Netherlands 10. FoldIt@Netherlands 14 pts. 12,747

  1. Avatar for pauldunn
    1. pauldunn Lv 1
    100 pts. 12,842
  2. Avatar for goastano 2. goastano Lv 1 99 pts. 12,839
  3. Avatar for Museka 3. Museka Lv 1 97 pts. 12,831
  4. Avatar for Crossed Sticks 4. Crossed Sticks Lv 1 95 pts. 12,830
  5. Avatar for actiasluna 5. actiasluna Lv 1 94 pts. 12,830
  6. Avatar for spvincent 6. spvincent Lv 1 92 pts. 12,824
  7. Avatar for Bruno Kestemont 7. Bruno Kestemont Lv 1 90 pts. 12,817
  8. Avatar for mirp 8. mirp Lv 1 89 pts. 12,815
  9. Avatar for Galaxie 9. Galaxie Lv 1 87 pts. 12,808
  10. Avatar for grogar7 10. grogar7 Lv 1 86 pts. 12,802


gitwut Lv 1

There is no filter for the hydrogen bond score. Is it a part of the Rosetta scoring function? How do we know if there are any h-bonds present?

Susume Lv 1

When I delete residues near 187 and re-insert them between 178 and 179, they become non-mutable. As a workaround, inserting them between 179 and 180 lets them remain mutable.

bkoep Staff Lv 1

Yes, the basic Rosetta scoring function includes several terms for hydrogen bonding. In this puzzle, we've adjusted the score function so that backbone-backbone hydrogen bonds make a greater contribution to the total score. This all takes place behind the scenes, so players don't have to worry about any extra filters; but note that top-scoring solutions will probably include many hydrogen bonds between backbone atoms (such as those found in α-helices and β-sheets)!

Players can view all hydrogen bonds by checking "Show bonds" in the View menu. The hydrogen bonding contribution from each residue is indicated by the "Bonding" sub-score (press TAB while mousing-over any residue to see the Segment Information dialogue).

Bruno Kestemont Lv 1

Just a question. If I calculate the subscores for the native virus protein itself only (segments 1 to 177), I get the following totals:

Clashing: -29355
Sidechain: -5417
Hiding: -4948
Reference: -706
Ideality: -705

Is it normal that a native protein scores so low ?

v_mulligan Staff Lv 1

Sorry for the slow reply – I don't want you to think we're ignoring this issue. I'm not quite sure yet whether this is a Foldit bug or a mistake made in puzzle setup, but it is something that we missed when test-playing the puzzle. We'll try to have it fixed for the next puzzle; in the mean time, please use the workaround.

v_mulligan Staff Lv 1

Yes. The scoring is based on the Rosetta energy function, which is inherently "rugged", meaning that very small changes in protein structure can result in huge changes in scoring. (A tiny clash can create a big penalty, for example.) When we import a crystal structure, the protein conformation is usually very close to, but not quite exactly the same as, a conformation that the scoring function "likes". We have two options in this case: we can relax the structure (basically, apply the "wiggle" function) to relieve very minor issues that the scoring function thinks are clashes, which will improve the score greatly but move the structure slightly away from the crystal structure. Alternatively, we can keep the crystal structure despite its non-ideal scoring. Since the scoring function is based on approximations, the first option does not result in a "better" structure; on the contrary, although there is some resolution limit to the accuracy of the experimentally-derived structure, it is almost certainly more accurate than the structure that's relaxed based on our imperfect scoring function, so we don't like to do that. So we do the latter, keeping the unrelaxed structure, despite the structural issues that the scoring function thinks are present.

To correct for this in Foldit, we add a constant bonus to the puzzle score so that players aren't penalized for the fact that the scoring function thinks that the target isn't quite right.

bkoep Staff Lv 1

Residues PHE 181 and TYR 182 are not allowed to mutate in this puzzle, and they should be colored dark orange.

If you insert a residue before or after an unmutable residue, the inserted residue may also be unmutable (see Susume's comment above). If you want to insert a residue between a mutable residue and an unmutable residue, be sure to select the mutable residue before using the Insert tool.

Bruno Kestemont Lv 1

Very interesting ! Thanks to make it understandable for us !

I wonder how the crystal structure is known.

On de novo puzzles, May be Foldit can give a proxy of the crystal structure, then further refinements by biologists are still needed I suppose?