Crowdsourcing and Distributed Computing

Increasingly, scientists are taking advantage of mass collaborations to cheaply generate ideas and bring a wide variety of perspectives to bear on research questions. Foldit is a computer-aided form of such crowdsourcing, but crowds can also aid computers. For example, in distributed computing, people volunteer their computer's idle time to be used in solving a problem. Individually, these processing cycles don't count for much, but combined, they add up to a virtual supercomputer. Made famous by the search for extraterrestrial intelligence's SETI@home program, distributed computing helps model protein structures as well. Rosetta@home, also developed at University of Washington, has been installed on hundreds of thousands of host machines, providing a valuable alternative to more traditional protein analysis techniques, such as X-ray crystallography and nuclear magnetic resonance spectroscopy (NMR).

Molecular Clockwork

An essential part of the watchmaker's craft consists of assembling a collection of delicate parts within as compact a space as possible, while ensuring that the tight quarters don't interfere with the timepiece's function.

In Foldit, players use a simple box of tools to manipulate the shape of a protein. The idea is to bend, twist, move and shake the protein's side chains and amino acid backbones such that the whole structure is packed into its optimum shape. Players know their solution works when they get rid of collisions between side chains of atoms, hide the hydrophobic chains inside the protein, face the hydrophilic chains outward and remove large empty spaces that threaten the stability of the protein -- all of which is reflected in their score.

The score, along with rules governing permissible moves, derives from the laws of physics governing protein folding. Thermodynamics tells us that natural systems tend toward states of lower energy. Other physical laws, such as the mutual attraction of opposite charges, repulsion of like charges and limitations regarding how atomic bonds can be arranged and rotated, are also built in.

The Foldit program abstracts the details into a form that the eye can perceive and the brain can grasp. Physics are handled behind the scenes, freeing players to manipulate the shapes via meticulous analysis, gut instinct or whatever method suits them.

Within a year of its introduction, Foldit players produced protein-folding solutions that outshone those submitted by molecular biologists. Inspired by early successes, Foldit's creators applied the program to other proteins and tasked players with designing new proteins to fight cancer, AIDS and Alzheimer's disease. For example, the p53 tumor suppressor protein is damaged in many cancer patients. If repaired or replaced, such a protein might stop tumor growth.

Successfully puzzling out the protease enzyme MPMV is the capstone of Foldit's career so far. Before they got there though, players churned through tens of thousands of ever-improving prototypes until, less than three weeks after they began, they had solved this particular protein puzzle [source: Niemeyer]. It wasn't a cure for HIV but, thanks to a retroviral family resemblance, MPMV's protein catalyst will help researchers build better antiretroviral drugs with which to fight HIV.

Foldit is not without its limitations, nor is it a Rosetta Stone for all proteins. Nevertheless, it enabled a crowd of players to predict the structure of a protein that had defied all traditional approaches, and that alone justifies its value as a tool for molecular analysis.