Redesigning IL-7R Binders

Hi Foldit players! We need your help redesigning protein binders!

I'm bcov, a graduate student in the Baker Lab. My PhD project is to make proteins that stick to other proteins. In my work, I’m given the model of a natural target protein and my task is to design a new protein that will bind to it. It turns out this problem is really hard because not only do my designed proteins need to bind to the target, but they have to properly fold first! Fortunately, I can use a high-throughput binding experiment that allows me to test 100,000 different proteins at once.

At the moment, I’m interested in studying the folding aspect of this problem. I have a clever experiment planned where I should be able to confirm the atomic accuracy of a designed protein even when it’s mixed with thousands of other proteins. For this experiment, I will need lots of binder designs that have different folds, but that share a common binding interface. I'm planning a series of Foldit puzzles in which players can redesign my binders while preserving the binding interface.

My designed binders target a protein called interleukin 7 receptor (IL-7R), which helps to regulate the human immune system, and is an important target for cancer therapy.

Here are the details of the experiment:
· I have 11 designed proteins that are confirmed to bind the target IL-7R
· I want to leave my designed binding interface the same, but redesign the rest of the protein
· In each puzzle, your task is to design the rest of the protein so that it folds the interface-side in precisely the right conformation
· Your designs will be tested for binding against IL-7R
· You will get a binding score based on how well your design binds in the wet lab

The binding score here is really cool actually. After we run the binding experiments at the end of the puzzle series, you will receive cold-hard data from the biochemistry lab about the binding strength of your design. Well-folded proteins that fold precisely into the puzzle structure will likely score the highest. Details about the binding score will be released later, but in general, there are three categories:

1. Your design did not bind to IL-7R
2. Your design bound to IL-7R but was worse than my design
3. Your design bound to IL-7R and was better than my design

If you end up in category 3, congrats! You beat me :P

Nearly all Foldit player designs will be tested experimentally. This is possible because we can test all the designs at the same time in our high-throughput binding experiment. Designs that look especially good will be tested multiple times with various mutations to increase data consistency.

Due to time constraints, puzzles in this series will be shorter than our normal week-long puzzles, and will only run for 4 days at a time. We'd like to generate as many variants as possible for the original 11 binders. So, don't worry if you miss a puzzle; there will be plenty more to follow up!

Check out the first puzzle of the series, Puzzle 1704: IL-7R Binder Redesign: Round 1, which is out now! Happy folding!

Update (8/16/2019): Read the followup post Protein Design Critique: IL-7R Binder Redesign

( Posted by  bcov 69 1821  |  Thu, 07/25/2019 - 20:37  |  5 comments )
Joined: 09/24/2012
Groups: Go Science

I'm wondering:

-is it still necessary to share to scientists? or do you pick up anything is sent to the server?

bkoep's picture
User offline. Last seen 5 hours 19 min ago. Offline
Joined: 11/15/2012
Groups: Foldit Staff
Yes, please share with scientists!

We do save all the models that are sent to the server; if you're playing Foldit online, then this includes autosaves and works-in-progress.

However, this means that we can have 100,000s of solutions at the end of the puzzle, and it's not always easy to sift through them to find the "finished" models. (Many players have multiple tracks in a single puzzle and produce multiple solutions, so it's not as simple as picking the best-scoring model from each player.)

Sharing a solution with scientists makes that solution stand out from the 100,000s of others, and we always look at the Share with Scientist solutions!

Joined: 05/19/2017
Groups: None
Experimental testing time

How soon can we expect "cold-hard data" once the results are in? I'm all for seeing what the game can generate but sometimes labwork takes time to do and communicate, such as the runtime of the Aflatoxin series.

bkoep's picture
User offline. Last seen 5 hours 19 min ago. Offline
Joined: 11/15/2012
Groups: Foldit Staff
That's a fair concern!

Once all the player designs are in, it will take 4-6 weeks to synthesize the DNA genes for all of the proteins, and then another couple of weeks to run the binding experiment and process the data.

In the best case scenario, we could have data within 8 weeks of the final puzzle—but no science experiment ever goes according to plan, so I can't say for certain!

However, unlike the aflatoxin experiments, which are carried out independently at UC Davis, I see bcov in the lab every day, so I can personally pester him for updates!

User login
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.
Social Media

Only search
Other Games: Mozak
Recommend Foldit
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, Boehringer Ingelheim, RosettaCommons