Thank you, Mario, but our protein is in another configuration!
By Thoreau
University of Washington researchers have developed a computer game that folds proteins. Yes, really.
Protein folding is a tough problem. You’ve got this long chain of amino acids, and somehow they come together to take on a particular shape. And they do this every time a cell produces the protein. You’d think that a long chain of molecules would try all sorts of random configurations, but somehow they always form the same pattern. It’s hard to understand why, because all of the molecules are interacting with each other in very complicated ways, so you have to solve an equation that could involve hundreds or thousands of forces. Some computers are having luck folding a few proteins, but it’s slow.
The researchers theorized that people are very good at playing games and picking out the most efficient way to do something. Good gamers intuitively know which traps to avoid. So they designed a game where you try to fold up a molecule into the most stable shape (the one where the forces all cancel and the energy is at a minimum). They tested it on some proteins with known structures, and they got good results. Now they’re trying proteins with unknown structures.
So to all my students, I say “Go and play more video games! Do something useful for science!”
I predict that we’ll start seeing research article titles like “PWNAGE of Matrix Metalloprotease 9 Structure! SW33T!” by DoneGeneMaster11 et. al.
Comment by Doug T —
May 9, 2008 @ 8:21 pm
It’s been 14 years (crap, I’m old) since I took my one and only biophysics class, but I thought you could generally get a good idea about protein structures from x-ray scattering, high res MRI, and other tools. (Which is how they got the known structures of the proteins mentioned.) What’s the problem with the unknown ones?
Also, knowing enough to be dangerous, I’d have thought this class of problems was a great candidate for a simulated annealing theoretical approach. Is it just too complicated for that?
To be further annoying, is it really known that proteins are folded into their lowest energy state? I’d think there’d be some path dependance, since they’re assembled one amino acid at a time, in a specific sequence, so that you’d likely end up in some sort of local minimum rather than the global min.
Comment by mds —
May 9, 2008 @ 9:28 pm
Sure. But it turns out to be difficult to crystallize many proteins in a way that yields good data from x-ray diffraction, or to make them soluble enough at high enough concentration for NMR. Despite all the confident statements from groups that do large-scale screening of crystallization conditions, it can still take years to get enough good data on a system of interest. So it would be really handy if we could just take the sequence and get the structure.
Nope. In fact, they probably are in a “sufficient” local minimum. The experimental methods you mentioned actually rely upon this, since there is still an energy minimization process involved in getting to the final, published structure. All we can aim for is a minimum that’s “close enough.” Note that this is tied up with your inquiry about simulated annealing. Even when constrained by experimental diffraction data, the space of possible solutions is too large to explore exhaustively via simulated annealing. Without experimental constraints? Forget it. For now, anyway.
Comment by Glaivester —
May 9, 2008 @ 10:06 pm
To be further annoying, is it really known that proteins are folded into their lowest energy state?
Well, prions are an example that shows that proteins are not always folded into their lowest energy state by the body, as the normal form of the protein is higher-energy than the prion form (which is why prions can catalyze other proteins of the same type to convert them into the prion form; this transformation is spontaneous and only requires catalyzation).
Comment by Thoreau —
May 10, 2008 @ 12:30 pm
No love for the thread title?
Comment by mds —
May 10, 2008 @ 6:33 pm
[Rolls eyes] Yes, Thoreau, it was very nice. But I think more people are familiar with structural biology techniques than Super Mario Brothers. Perhaps next time something along the lines of “Diffracts of Life” or “hepA Goes to Monte Carlo.”
Comment by von Laue —
May 10, 2008 @ 10:24 pm
diffracts of life — oh, my. Do you have more of those?
Comment by Thoreau —
May 10, 2008 @ 10:37 pm
CryoTEM me a river.
Comment by socratic_me —
May 12, 2008 @ 10:43 am
I thought the title was pure genius, but not nearly as cool as your suggest future article title.