With computer games that challenge citizen scientists to design protein and RNA strands, Adrien Treuille might have changed the way researchers think about collaboration. His games, Foldit and EteRNA, are played by thousands.
Treuille, a PopTech fellow and assistant professor of computer science and robotics at Carnegie Mellon University, told me in a recent interview that when it comes to protein folding problems, untrained online gamers are more talented than even the best computer programs. Below are excerpts from our interview.
What made you turn to computer games to help people understand movement, shapes and structures?
Computers let you make worlds appear out of thin air. When I started doing computer science, I was immediately drawn to graphics and simulation. With these games, it's like taking that idea to the next level. If you can create a program that simulates a protein or an RNA, that means people could experience these things. They could almost be the same size as an RNA and fold it like a toy. If you can play with it like a toy, maybe you can figure out stuff no one else knows.
You're basically crowd-sourcing to find answers to scientific problems. How successful has this been?
We published a paper in Nature that showed people were better than the best computer programs at solving protein folding problems. In the case of RNA folding, the difference is even more dramatic. Our best models are not very good at predicting how RNAs are going to fold. We're allowing people to do what humans do best, which is to find the patterns.
The key to the game is it's experimental. When you're playing, you can click the "submit" button and the RNA that you designed will be synthesized. You'll get a picture on your screen of what it looks like in real life. We have basically 25,000 untrained experimental scientists kicking ass at RNA design. It's fair to say that some of the best empirical RNA designers in the world are our players.
What are the real-world implications for this work?
This can and will go in so many different directions. The elite players have become sophisticated scientists. If I had enough money to hire our top players in my lab, I would. I went through all this scientific training and finally ended up at Carnegie Mellon. But many people are as talented or smarter than me and never got a crack at science because it's so daunting to get into. One of the many trends we see in these games is they provide a forum for people to do what they're good at in ways the traditional labor market hasn't allowed.
Why is this work important to you?
I think we're going to make real medical advances based on what the players are doing. That's exciting.
We couldn't do the kind of data analysis we do without 25,000 people looking at the data. This is science on a massive scale that was never possible before. It deeply affects everyone, especially the players. It changes the way they look at the world and their own capabilities. It also changes us. In academia we try hard to raise money for one or two grad students. Through these games, we now have access to a pool of 25,000 people. It changes the way we look at what types of scientific problems are possible.
What do you mean by medical advances?
Protein folding and RNA design are not currently the basis for drugs we take. But they are the basis for how the body works. They're like the drugs inside the body that make it operate. In all these cases, the goal is to understand how these devices work and build new things that mimic the way the body works. For example, in the case of RNAs, they play a deep role in cancer and AIDS. Hopefully by building drugs which use the same substrate as these diseases, we find novel ways of attacking them.
Is your work part of a trend?
The biggest trend is the Internet. At the largest scale, the Internet is allowing collaborations of millions of people in a way we've never experienced. As part of that broader trend, we're trying to bring it to complicated experimental science. Within that, there's the smaller trend of citizen science. It's using the Internet to coordinate scientific projects. Within those trends, EteRNA is unique in the complexity of the problems we're asking people to solve. Also, it's based on real-time experimental science.
Were you surprised by the talent of the citizen scientists?
Yes. But I'm getting less surprised with time. This was the second game I made. Foldit was the first. Both times, the outpouring of obsessive game playing boggled my mind. There's lots of enthusiasm. When you put things out there, you can connect to amazing subcultures. It is astonishing to me, but maybe we should expect that the Internet can allow us to find specialized subgroups.
What's next for this work?
We're coming up with several new games that explore protein and RNA and drug design. We're trying to understand from a theoretical property why these games work and how we can help other people create games that will benefit them and benefit the labor market.
On the experimental side, we're looking into how we can do a lot more experimental science on the Internet. Why can't we have a World Wide Web of experiments where at the click of a button an experiment is done?
Photo: Adrien Treuille