This was the puzzle that we posted for this target during CASP10:
Here is the sequence logo predicted by the SAM server:
H = helix
E = sheet
C = loop (or coil)
The taller the letter at each position, the higher the probability of that specific secondary structure for that amino acid.
For example, the Valines at residues 37 & 45 are highly predicted to be part of a sheet. However, the Valine at residue 138 is not predicted to be anything!
Q is the keyboard key to put your protein on the middle of the screen.
I just notice that the provided Electron density cloud is shaped like a box/cube or more precisely a parallelogram.
Is that intentional and/or an artifact of the scanning process?
That's a good question, bertro.
The box shape of the electron density cloud is an artifact of the scanning process, as you've guessed.
I believe getting an electron density goes a little something like this: we crystallize many copies of the protein. Then we use x-rays to scan these many crystallized copies. These copies are 'tiled' according to the crystallization pattern, and thus, the density we get is self repeating as well. Often times there are no clear boundaries to where one protein in the crystal ends, and another begins.
So what we've given you is just one 'tiled' unit of the measured density.
When you give us one "tiled" unit does it means we could have just a part of the protein on the left and the rest on the right for instance? Like the protein being split in x parts or is it all contained as a whole in the density map? Could that throw off the "Align Protein to Density" results? Not even thinking about how we perceive the protein to be like in space... if that is the case.
What I mean by 'tiled' is that there are many identical copies that are sort of 'stacked' symmetrically (ONLY translation, no rotation).
Because of this, where we actually start and stop drawing the boundary of the density visualization can catch a piece of one copy of a protein, and a different piece from another copy.
The protein is all one piece, but the actual visual density may show parts of the NEXT copy of the protein.
You'll notice that your density score still changes, even when the entire density visualization is outside of the 'cube'. This is because the density repeats itself in the tiling pattern mentioned above.
That's a good point about the "Align Protein to Density" button - it likely will not function quite right in this puzzle.
You may be thinking, "Why didn't they just center the box on the protein?" There may be no box that holds just one protein molecule, it can stick out. M.C. Escher became famous doing this sort of thing.
Here is a neat flash app to make a simple tile. The box is defined by 3 points (2 lines). If you never move those points, the underlying box never changes. But you can arbitrarily reshape the edges into anything and it still tiles, based on that original box, even if the tile isn't remotely box shaped.
If you took 8 of those ED boxes, and stacked them 2x2x2, you'd have one complete ED in the center. You should have some protein sticking out of the box, which would fit on the other side.(*)
Some people can't find the ED. It's way off in space. When you first start the puzzle, hit "Align to Density" and then Q.
jflat says the scoring function repeats, although the drawing doesn't. That means wiggle will still try to align to the protein. You can work inside the ED, and when you finish, you can freeze it, and use the move tool to move it to the opposite side of the ED. Wiggle will then align your protein to the ED, back the way it was (if you were close to correct). The part that was inside is sticking out, and the part that was sticking out is inside.
(*) Maybe we could try one with the larger cell, see if people prefer that.
Thanks for the tips brow. I'll add the bit about getting the protein on top of the density to the description (usually this isn't an issue, I just forgot to save after moving it when I posted the puzzle).
It's probably accurate to think of the proteins that are shown in the density as tessellated. They don't quite fit together perfectly like a tessellation, but where one ends, another begins, and they all have the same orientation - they're just translated copies.
How did you decide where to cut the ED?
That's a good question, Susume.
The density for this puzzle was supposed to contain one WHOLE copy of the protein, along with pieces of other copies. Unfortunately, after doing some analysis, it looks like the entire copy wasn't contained in the density provided. For the next round of this puzzle, we'll have the complete density.
Sorry about the incomplete density - as you can tell, it isn't always obvious what's going on in the density, so it's hard to tell when something looks off. We're also working on some more options to make working with the density easier, and we're hopeful with these changes that players will be able to solve the experimental densities as well as the ones we've tested so far!