The work from the University of Washington demonstrates the power that modern computer simulations have in overcoming some potential shortcomings in regular methods of solving a protein's structure. Through the sophistication of a program such as Foldit, combined with the crowd-sourcing power of the Internet, the hunt for protein structures now has a new tool with ever-growing effectiveness and potential reliability.
However, if you are not as familiar with this field of science, then you might be wondering what scientists actually do to get the structure of a protein. Obviously many proteins, including this HIV protease, are not at all visible to the human eye, or a microscope, or a scanning electron microscope, or anything that allows you actually "see" a protein and how it folds. Through a somewhat indirect method of visualizing a protein, scientists have reliably solved many protein structures through x-ray crystallography.
As you might be able to glean from the name, this technique involves the bombardment of x-rays onto a crystal of proteins. At a high enough concentration when dissolved in a solution of water and other chemicals, proteins can fall out of solution and form crystals. When these crystals are big enough, they can be harvested from the solution and mounted on an x-ray beam.
While shooting the crystal with x-rays, scientists collect diffraction data from a detector that is mounted behind the crystal, looking to see where the x-rays scatter as they pass through the crystal. If an x-ray hits something as it passes through the crystal, such as an electron, it will scatter off at an angle from its original path straight down the center. For those of you with a little knowledge of chemistry, this is similar to Rutherford's gold foil experiment, where he shot a thin layer of gold foil with alpha particles and determined that an atom has most of its mass concentrated at its nucleus.
Of course, x-ray diffraction data gives scientists a much more complex picture, collecting an extensive data set from shooting the crystal over a range of angles. While I myself cannot explain the complex calculations involved, refinement and processing of the diffraction data allows scientists to generate an electron map of the protein. With this map, scientists can build a model for the protein based on its amino acid sequence, and with a little luck and persistence the structure of the protein is solved.
As with any experiment that involves so many steps, solving the structure of a protein has many pitfalls as well. Proteins can refuse to crystallize, form terrible crystals that are far from being organized in any sensible pattern, fail to diffract x-rays, or give researchers a terrible data set to work with and refine. Flexible portions of proteins may also fail to stay still and diffract x-rays properly, leaving a gap in the structure of the protein.
Other times scientists will have trouble generating a sensible model using the electron density map. They can get a rough idea for some portions of the protein, but other parts may not make any sense based on their experiments. It is here that Foldit may pick up the slack and help scientists out of this little rut with the help of a vast resource of human thought processes and reasoning.
One need not be an expert in protein structures or the physics involved to partake in the puzzles presented by Foldit. Built into the program are some primers on some of the interactions within a protein that determines its structure and a scoring system based on how realistic their model is based on the physics of protein folding. With both, gamers can set about to find the best way for a protein to fold in the absence of complete or reliable electron maps, doing their best to solve a puzzle with a little guidance from the computer's calculations. Harnessing the reasoning skills of humans that computers lack can yield splendid results such as mystery of the HIV protease's structure.
As is the case with this protease, solving the structure of a protein can lead to further experiments looking for other compounds that can bind or interact specifically with the protein in question. This sort of work can lead to drugs that are much more accurate at binding to or inhibiting the function of certain proteins while minimizing the chances of side effects.
