When Gamers Beat Scientists at Their Own Game
Video games sometimes get a bad rap. So it’s nice to see some unanimously good press surrounding the subject.
A few weeks ago, a paper was published in Nature, the most accurate findings of which were found by a 476 person strong group of gamers. Using a game called Foldit, they discovered the exact shape a protein with more accurate results than the research group that assigned them the task, and uncovered a new group of proteins that could potentially help in the fight against Alzheimers. Pretty good going for an afternoon hobby.
At first it may seem a bit of a misnomer to say that the protein is ‘discovered’ as if you already know what it looks like and know what you’re looking for, it’s pretty hard to see at first how this can be construed a discovery. The thing to understand though, is what was already known and what the researchers were looking for.
What was already known was what is called the electron density map of the protein. This is effectively an ‘outline’ of the protein as by using an electron microscope, you can build up a 3D image of where the electrons of the atoms in the protein are. However, at this scale, the wave like nature of the electron comes into play and the exact location of the electron becomes unclear, meaning there will only be a certain probability you see it in a particular place. In a sense, this probability is what is measured by the microscope, as the higher the probability of finding the electron in a given place, the larger the reading on the microscope. Therefore the actual shape of the molecule becomes unclear, as the folding and branching of the protein cannot be seen at this level of low detail. Much like trying to deduce the exact appearance of a dog by looking at a balloon animal (of a dog, I think it would be unreasonable to guess what a dog looks like by seeing a balloon snake).
What they were looking for was this increased level of detail, to understand how the amino acids within the protein were ordered and arranged in 3D space. This was done by comparing the electron density map of the gamer’s molecules with that which were experimentally found. For the gamers, the closer the designed structure was to the original, the higher the score achieved.
The reason that the games won here was because the 476 strong group managed to produce a protein structure that fit the electron density map better than the team of scientists containing 2 experts in the field . Which to me, gives an interesting look at how, for certain problems, crowdsourcing science can produce very useful results. It has already been happening with entire teams dedicated to different projects, such as the Zooniverse website.
Now this story does come with a bit of a disclaimer, as it is essential to remember that without the guidance of a professional with a well grounded knowledge of the research area, none of these findings could have been obtained. Here, it was entirely necessary to have researchers dedicated in a professional capacity. As without them, there wouldn’t have been the electron density maps, the knowledge of the protein structure up to that point, or even the game to play.
That being said, these sorts of studies do seem to give the impression that in the future, some problems may best be solved not by a small group of experts in it for the prestige. But instead, if given the right direction, by a large group of hobbyists in it for the fun.
Max
Sources:
quantumwalrus 