harnessing the power of the gamer

There are many things in the world of science that we just can't figure out. Not because we don't have the math, or the theory down but because we simply don't have the computing power available. It's a well known fact in my lab that I'm a bit of a video game geek, my boss regularly jokes that if they could some how find a way to make downloading research papers be controlled by a dualshock and have a snazzy graphical interface (possibly with a J-pop soundtrack) I'd probably never leave the office.

He also thinks this is the key to solving most problems of computation: to code the problem into a video game and have the processes required to solve it be related to actions taken within it. He's talking "massive scale, solving the problems of the universe which are possibly not for human consumption" that would require large amounts of obfuscation before being turned over into the hands of gamers of course, so there's no way we could implement it in the near future. But it's nice to see that a program named FoldIt has actually taken the first steps.

The game involves trying to figure out different ways in which proteins can be folded. As a biochemist this is a fascinating topic for me, protein folding is ludicrously complex and defined by a whole host of factors: the order of amino acids making up the primary chain, the secondary structure that the polypeptide takes, stabilising interactions between individual residues, the points at which water interacts with the molecule (I won't go on but if you're interested the Wikipedia page on protein structure acts as a fairly good crash course). Calculating all the different possible configurations is a mammoth task, but using human interaction could make it solvable much, much quicker.

Right now they're testing out the accuracy and efficiency of the game using known protein configurations but if it works well enough we're not just talking about solving the structures of existing proteins but also determining novel structures. One of the biggest problems facing drug design has long been the inability to isolate the perfect protein structure for the desired function. This is especially important in treating viral diseases such as HIV where interaction between the drug molecule and the virus' structure is absolutely vital to correct function. Computers so far aren't capable of determining completely novel protein structures from scratch and so if this technology actually works (and is fun enough to convince people to play with it), we could certainly see protein science, if not medicine advancing in leaps and bounds.

More information is over at Science Daily and Kotaku. I don't know about you guys but solving 3D puzzles and helping humanity? Sounds pretty spiffy to me.