Alejandro and I sat down for an interview with Seth Cooper, the Creative Director of the Center for Game Science at the University of Washington. Dr. Cooper spoke during TEDMED about the power of play and the role that games can play in dealing with scientific challenges. Seth is a co-creator and a primary developer of Foldit, a protein folding game that allows users to help predict protein structure and design new proteins, both of which allow for the better understanding of disease and also for the development of new ways to combat disease.
CJ: In your speech today you talked about Foldit and generating the best protein shape. How is that best shaped protein determined exactly?
Seth: Within the game, everything is based on a score function obtained from biochemists. The game is based on a computational function, so if you give it a protein structure, it will give you back a number that tells you how well folded that particular protein is. It‘s based on the fundamental properties of proteins. It’s an estimate of how well folded that particular protein structure is. The way that the game works is that we post a protein structure and the players download it on to their computer, and they can manipulate it and move pieces around. They are trying to fold the protein up in the most compact possible way. And while they do that, they can see their score change in real time; they can see if they are making the protein better or worse. The goal in the game is to get the highest score, which means you have the best-folded protein structure for that particular protein. And, it isn’t all just competition. Players can form groups and develop specialized skills for different parts of the protein folding process.
Alejandro: Do you have to know anything about how proteins fold to play the game or can anyone play?
Seth: You can be anyone. It’s all about the shape of the structure; it’s like a 3D puzzle. At the heart it’s about fitting pieces together. There are particular protein folding rules and those are all taught to the players in a series of introductory levels that we have. There are 25 or so of them now. They start off simple where a protein has something wrong with it, and the game tells you how you would fix that particular problem. When you fix the protein you go onto the next level, and it gets more complicated from there.
CJ: One of your recent discoveries from Foldit was a protein involved with AIDS. Could you explain a little more about what that was specifically?
Seth: This was last fall when we had the paper come out. The particular protein that was of interest in that project was the Mason-Pfizer Monkey Virus Retroviral Protease, which is a key protein in this virus that leads to AIDS in rhesus monkeys. We were collaborating with some scientists in Poland who had a fair amount of data on that protein, but they weren’t quite able to solve for what the actual shape of it was. Basically, they had enough data so that if someone gave them the shape of the protein, they would be able to say this fits the data, and that’s right. But, the scientists didn’t have enough data to actually get the shape. We took some guesses at what the structure might be like, but were wrong, and so we gave them to the players of Foldit for about three weeks. And then, when the data came back at the end from the players of the game, the biochemists looked at it, analyzed it, and found some structures that fit the data that they had. They could then say that this was the right structure for that particular protein. The cool thing is that now that we know the shape of that protein we can try to design another protein or a drug that would prevent the protein from folding properly, thus effectively deactivating it. We could make a drug to fight against the disease.
CJ: Are you focusing primarily on Foldit right now or you working on some other projects too?
Seth: At the Center for Game Science where we developed Foldit, we are continuing to develop Foldit and look at new ways to use that to solve different problems. We are looking at flu inhibitors now and are having players design some of those. We are also developing new games and are looking specifically at early math education. We are trying to develop a whole world of different games that can, in a sense, turn education into data driven science. We do this by gathering a lot of data on how people play the games, how they get stuck, how they can get unstuck, and how they get confused. This is done by having lots of people interacting with the games.
CJ: When did you start looking at flu inhibitors? Is this a more recent development?
Seth: That’s relatively recent actually, but we have been looking at it for a couple of months now. But, we will really start to ramp up the interaction with the players on that particular series of puzzles. These synthetic proteins are really cool. We recently had one synthetic protein that the players helped design; it was an enzyme that catalyzed a particular organic chemistry reaction that some biochemists were interested in. There was a ton of back and forth on a series of puzzles between the players of Foldit and scientists. What the scientists can do is take the virtual protein that the players designed in the game, test it in the wet lab, and actually run experiments and end up with a real synthetic protein that never existed before in nature. A player can make a protein in a game, and the scientists can then make it in the real world. They can see if it works, if it folds up at all, and if it does what we are interested in. We just had a paper come out going through this process for a particular enzyme, and we ended up with an enzyme that was significantly more efficient than the one the scientists had started out with. The players made a drastic change to the starting protein and added a huge piece to the protein, and it turned out to work really well. We are looking to follow this same process but with something that applies to fighting against the flu.