Placeholder image of a protein
Icon representing a puzzle

2062: VHL ligand design: Round 2

Closed since 11 months ago

Intermediate Overall Small Molecule Design


October 27, 2021
Max points

Design a small molecule which can bind VHL! This is the second in a series of puzzles using the small molecule design tools to design small molecule binders to the von Hippel-Lindau E3 ubiquitin ligase. We don't want you to modify the protein, but instead we'd like you to build new small molecules which fit into the VHL pocket. See the new "Inside the Pocket" and "Fun with Fragments!" tutorials to learn how to use the small molecule design tools.

Note: To get the most out of the small molecule design tools, we recommend changing you view settings:

  • Set Color: Score/Hydro+CPK

  • Set View Protein: Binding Site or Cartoon Ligand

  • Turn on: Show Clashes, Show Voids, Show bonds (non-protein), Show bondable atoms

  • Turn off: Show exposed, Show bonds (sheet), Show bonds (helix), Show bonds (loop), Show bonds (sidechain)

For this round, we're encouraging you to do more to change the core of the compound, specifically the central proline-like five-member ring (pyrrolidine). Changing that group should get you a bonus.

We've also adjusted the objectives to address some issues we saw in round 1. In particular, we've added penalties for enol and enamine groups, which rapidly rearrange in cells. We've also penalized thiocarbonyls. These join the penalty for amide groups and the pyrrolidine in the "Bad Groups" objective. We've also added a cLogP objective to keep the compounds from getting too "greasy", which is bad for drug properties. You'll need to carefully balance removing polar groups to get the TPSA bonus, while not making things too hydrophobic and losing the cLogP bonus.

This is a project in collaboration with Boehringer Ingelheim. Boehringer Ingelheim has committed to help evaluate and test the molecules which Foldit players have designed. All compounds created as part of the collaboration puzzles will be made publicly available. Experimental results from testing the molecules will also be released publicly. All participants and game sponsors of current and future small molecule design games commit to complying with the Foldit Terms of Service including those pertaining to intellectual property.

Top groups

  1. Avatar for Contenders 100 pts. 20,111
  2. Avatar for Gargleblasters 2. Gargleblasters 73 pts. 20,080
  3. Avatar for Anthropic Dreams 3. Anthropic Dreams 52 pts. 20,080
  4. Avatar for Beta Folders 4. Beta Folders 36 pts. 19,952
  5. Avatar for Go Science 5. Go Science 24 pts. 19,906
  6. Avatar for Australia 6. Australia 16 pts. 19,894
  7. Avatar for L'Alliance Francophone 7. L'Alliance Francophone 10 pts. 19,854
  8. Avatar for Marvin's bunch 8. Marvin's bunch 6 pts. 19,817
  9. Avatar for Void Crushers 9. Void Crushers 4 pts. 19,810
  10. Avatar for Italiani Al Lavoro 10. Italiani Al Lavoro 2 pts. 19,762

  1. Avatar for Aubade01
    1. Aubade01 Lv 1
    100 pts. 20,366
  2. Avatar for Nicm25 2. Nicm25 Lv 1 95 pts. 20,303
  3. Avatar for fiendish_ghoul 3. fiendish_ghoul Lv 1 90 pts. 20,162
  4. Avatar for spvincent 4. spvincent Lv 1 85 pts. 20,111
  5. Avatar for Bletchley Park 5. Bletchley Park Lv 1 81 pts. 20,088
  6. Avatar for HuubR 6. HuubR Lv 1 76 pts. 20,080
  7. Avatar for robgee 7. robgee Lv 1 72 pts. 20,080
  8. Avatar for grogar7 8. grogar7 Lv 1 68 pts. 20,076
  9. Avatar for Galaxie 9. Galaxie Lv 1 64 pts. 20,024
  10. Avatar for BootsMcGraw 10. BootsMcGraw Lv 1 61 pts. 19,972


rmoretti Staff Lv 1

Information about the objectives in this puzzle. (All objectives give +1000 points when fully satisfied.)

  • Bad Groups - Gives a bonus for groups we're interested in removing, including amide, pyrrolidine, enol, enamine, and thiocarbonyl.
  • Ligand cLogP - A measure of polarity - Keeps the molecule from getting too hydrophobic
  • Ligand TPSA - Topological Polar Surface Area - Keeps the polar surface area (including buried polar surface) low.
  • Ligand Hydrogen Bond Donors - Keep the number of ligand hbond donors low.
  • Molecular Weight - Intended to keep the ligand from getting too big.
  • Number of Rotatable Bonds - Intended to keep the ligand from getting too big and floppy.
  • Ligand Substructure - Identify areas of the small molecule which aren't very drug-like.
  • Synthetic Accessibility - Identify molecules which are likely to be hard to make in the lab.
  • </ul> Most objectives have a "Show" button which should highlight areas of concern in red. (Even when fully satisfied, objectives may highlight groups which may eventually contribute.)

BootsMcGraw Lv 1

…is not to play.

No matter what is changed, it triggers one or more of the penalties and/or drops the score hundreds or thousands of points. Get rid of the amide… lose 300 points because cLogP is now too high. Add an oxygen to fix the cLogP… and get a Ligand Hydrogen Bond Donor penalty of 200 points. Add a nitrogen, instead… and get a thousand point penalty for having undesired groups. Why do you think half the players are still at the starting score of 19,760 points?

You've got to give us dummies some guidance on what to do, or we will walk away in frustration.

1/10… would not play, again.

BootsMcGraw Lv 1

… I managed to break past the starting score by literally making random changes, until something scored better. Not the best scientific approach. Is this what we want?

rmoretti Staff Lv 1

One of the reasons we aren't giving clear "do this to win" instructions is that we're not sure what is going to work well.

The objectives are based on real considerations – There's certain groups which cause issues with drug development, so we penalize them. We know that the starting structure is a bit too polar (and too like a peptide) for being a good candidate. On the other hand, if you get too hydrophobic the compounds aren't good either. A good small molecule design has to strike a fine balance between the two extremes. (Enough polar groups such that it's not too hydrophobic, but not too many polar groups, and definitely not certain types of polar groups.) We might be setting the objectives a bit too tight on this puzzle, though.

Keep in mind it's very early days for small molecule design in Foldit. We're still figuring out how to dial in the objectives to get good results. Basically, think of the state of protein folding in Foldit back in 2008, or the de novo protein design puzzles back in the early 2010s. It's a learning experience on both sides – on our side on how to set up the puzzles to get the sort of designs that are worth making in the wetlab, and on your side to figure out how to work with the puzzle setup to make compounds that score well (and are worth making in the wetlab).

The hope is that the community can come together to produce resources on how to approach small molecule design puzzles from a citizen science perspective. Crowdsourcing drug design isn't something that has a lot of precedence. Experts in big pharmaceutical companies spend a lot of time on similar sorts of problems, with a rather low success rate. There's some resources and institutional knowledge out there for experts, but currently not a lot targeted at citizen scientists.

The Foldit developers (and our collaborators at BI) are certainly willing to help out with information if we can, but we're also a bit in the dark as to what will be maximally useful for you. To some extent, you might want to view these early small molecule design puzzles as a challenge to figure out "how to play" more than one where you apply well-worn techniques. – There's a bunch of expertise and institutional knowledge that Foldit players have built up over the years for working with proteins. We don't have that (yet) for small molecule design, so we have to muck through together to build that up.

robgee Lv 1

from the 19760 base score, if you replace an atom with the same atom then wiggle your score goes up by a few points, helpful to get it started.

As to method, i just did the same as Boots, make a change, wiggle it, score goes up else reset, repeat.
Yeah i've got no idea what i'm doing and it is frustrating but thats kinda foldit in a nutshell :)

spvincent Lv 1

Right now it's pretty hit and miss. Mostly miss. A couple of things in particular:

As Boots has already mentioned, the size of the penalties is way too big (exception: the 5 point penalty for an amide). For example make the ligand just a little more hydrophobic by adding a random methyl group somewhere and the cLogP score can drop by 300 or so. In the context of the puzzle that might as well be a million, as there's no way a compensating improvement will be that big.

Also the random crashing continues. It's a little better now the crash on loading issue has been fixed but still the life expectancy of Foldit when playing this puzzle is less than 5 minutes. There doesn't seem anything reproducible about the crashes either. It's a pity because I like the idea of this kind of puzzle but having to restart Foldit all the time is just too tiresome.

cas0362 Lv 1

For this game tetra alkyl/aryl ammonium species should likely be avoided. Ideally basic amines in the suggested molecules should be also avoided as this will make it easier to install a basic centre in the protein of interest binder in the resulting PROTAC (if the target protein requires it).