Foldit Lab Report #2: Symmetric Design Tips

Happy Fall! Check out Foldit Lab Report #2 on YouTube now!

- Foldit has a new community liaison
- macOS update broke Foldit (but we fixed it)
- New opinion poll about data sharing

- IL-7R binders
- Aflatoxin enzymes
- Symmetry, symmetry, symmetry

- New genes are on their way to the lab

- bkoep shows what makes a designed interface look good

(Fri, 11/01/2019 - 02:00  |  10 comments)
Joined: 05/19/2017
Groups: None
Re: Symmetry Design

I haven't been doing these puzzles lately because I've been short on time and interest, but that point about interface residue choice confounded me a little. It's easy for Mutate, (and by extension Recipes) to choose hydrophobics for burial as is the metagame to do so. But sprinkling in those blue residues to avoid the folding problems described doesn't always contribute to score, and in the recipe "set it and forget it" metagame, those key decisions can be lost in the wash unless we know what we're looking for and freeze it. Are the limited interface puzzles your way of addressing this? Also what kinds of symmetry scaffolds should mid-level non veteran players (read: me) be trying out? I'd love to see more players recording their folding process and maybe even commentating over it so we can learn from each other's styles to save time and score better.

bkoep's picture
User offline. Last seen 3 hours 47 min ago. Offline
Joined: 11/15/2012
Groups: None

This is the crux of the problem we're facing. Foldit likes hydrophobic residues at these interface positions because they give a higher score (lower energy).

But the Foldit score isn't the whole story! The Foldit score doesn't consider alternative states of the protein (like unfolded states, or aggregated states). So when we go to make these proteins in the lab, instead of folding into the designed state (e.g. a symmetric trimer), they adopt some other state that would presumably have an even better Foldit score.

We thought the 'limited interface" approach might help with this, because these hydrophobics tend to appear at interface positions. Therefore, fewer interface positions should lead to fewer hydrophobic residues. It seems like this approach has helped things somewhat, but we'll have to see how these "limited interface" proteins behave in the lab!

Joined: 05/19/2017
Groups: None

Obviously there's no retooling Foldit's engine anytime soon to account for this, but I think that more can be done with bonus points rewarded with Objectives to compensate for this. It's going to be difficult to do the math on how much bonus points should be awarded to compensate for scoring losses, but it could definitely help in letting novel solutions that you'd be interested in rise to the top instead of getting "lost in the wash" of Recipe refinement.

bkoep's picture
User offline. Last seen 3 hours 47 min ago. Offline
Joined: 11/15/2012
Groups: None
On the nose!

This is our thinking exactly! We have plans for a new Objective that detects polar atoms without H-bonds (in fact, this is exactly what Baker Lab members use).

On our end, the implementation will be a little tricky, because this Objective will be very slow to run (we have to run expensive surface area calculations to figure out if a polar atom is buried, or if it can make H-bonds with the surrounding water). This means we will not be able to run the Objective in real time. Either it will need to be a "push-button" Objective that players have to run manually; or else it will have to run asynchronously and update itself every couple of seconds (this approach has its own problems).

spvincent's picture
User offline. Last seen 1 hour 9 min ago. Offline
Joined: 12/07/2007
Groups: Contenders
Thanks for the updates

Thanks for the updates bkoep.

I was wondering if the issue of too many hydrophobics in the monomer of a trimer could be addressed by resurrecting Hydrogen Bond Networks?

bkoep's picture
User offline. Last seen 3 hours 47 min ago. Offline
Joined: 11/15/2012
Groups: None

I think you're right, and we will probably see some more H-bond Network puzzles soon! The difficulty with the H-bond Networks was always that they tend to invite buried nonpolar atoms (the other big problem we're seeing). So I think we will have to be very strict about the "satisfaction" score of the networks.

Joined: 04/28/2015
Groups: Go Science
Thank you for the video. I am

Thank you for the video. I am engaged in the popularization of science, and people are asked to show how the creation of synthetic proteins occurs, how artificial genes are introduced into bacteria through, and how protein is extracted.
Most people simply do not understand what nanoworld is, and what size it is. fantasy and magic.

Since when it comes to complex sciences, people are distrustful of this topic since it cannot be seen or touched, and all microbiology and chemistry are based on complex devices, the principle of work that can only be understood by people with education or scientific degrees.

Joined: 02/15/2018
Groups: Go Science
Thank you for this video

Thank you for this video series and the recent updates.

It is really good to see how the Foldit team evaluates the players' designs (also thanks for the tips). I am wondering is there any plan for design puzzles that have more than three identical units like tetramers, pentamers and so on?

bkoep's picture
User offline. Last seen 3 hours 47 min ago. Offline
Joined: 11/15/2012
Groups: None
Higher symmetry in the future

Yes, we are absolutely interested in higher-order symmetries with more subunits! But our plan is to progress incrementally, since design tends to become more difficult with increasing numbers of subunits.

The symmetric dimers are a special case, because they can be accomplished with a single binding interface between the two subunits. Trimers, tetramers, etc. are more difficult because each subunit has to support two binding interfaces. Once we iron out trimer design, we expect tetramers to come relatively easily.

Download links:
  Windows    OSX    Linux  
(10.12 or later)

Are you new to Foldit? Click here.

Are you a student? Click here.

Are you an educator? Click here.
Only search
Other Games: Mozak
Recommend Foldit
User login
Top New Users

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, RosettaCommons