Sep 19 2011

Gamers Succeed Where Scientists Fail

The rather provocative title of this post refers to the at home science game, Foldit. The game essentially involves figuring out how to fold proteins in three dimensions. This is an extremely difficult problem, even for computers. The Foldit game is a way of harnessing the brain power of video game players to help find solutions.

The game has been available for a while, and now (apparently for the first time) has been used to solve a specific puzzle of protein structure. Scientists have been trying for 10 years to solve the structure of retroviral protease, a protein-cutting enzyme found in HIV-like viruses. So, they put the problem to Foldit gamers, and they solved the structure in one week.

The reason these types of problems are so difficult is that the number of possible ways in which a large protein can fold is staggering large. This is, fact, called an NP-problem (non-deterministic polynomial). The classic example of this is the traveling salesman who wishes to map the minimum route to a set of cities he wishes to visit. Such problems are impossible to solve by computational brute force (the number of possibilities to check quickly becomes greater than the number of atoms in the universe), and there is no mathematical way to derive the answer quickly.

For this reason computer programmers have simply not been able to develop a program that can solve the folding puzzle for large proteins.

The human brain, however, is a massive parallel processor and is excellent at pattern recognition – something we still do (for now) better than computers. Also, there are lots of people – lots of computer brains that can be put to work.

That is the idea of the Foldit program – to put human brains to work to solve the NP-problem of folding proteins that computers cannot do with brute force.  In this case, the Foldit gamers were able to suggest possible folding strategies that checked out – they essentially solved the folding puzzle in one week. The potential benefit of this is, potentially, the ability to design drugs that target receptor sites on the folded protein, perhaps to inactivate it. In other words, this knowledge could lead to the development of new anti-retroviral drugs.

It is important to recognize that the reason computers cannot solve NP-problems is not a limitation of current computer power. Standard computers, no matter how powerful they become, will simply not be able to solve NP-problems. However, some think that quantum computers may be designed to solve NP-problems (even if they are not particularly suited to running Windows).

The Foldit program is a great way to engage the public with science, and it’s a great way to make use of a powerful resource (the human brain) to solve difficult problems in science. It’s good to see the fruits of this program – hopefully it will make Foldit and programs like it more popular.

13 responses so far