3 replies [Last post]
Diderot's picture
User offline. Last seen 6 years 43 weeks ago. Offline
Joined: 05/09/2008

Mr. Fixit asked on an earlier thread how people get high scores. I've come in first in one puzzle and am often in the top 20 on others, so I guess I'll take a shot. Keep in mind that my suggestions are very, very preliminary: Everything I say here could easily be overthrown, and I'm far from the best folder around. Still, no one else suggested anything, so here I go.

What follows is a sort of loose algorithm: It's a set of procedures that you may, and perhaps should, deviate from at any time. In folding I may start or re-start from any one of these suggestions, or I may do something not listed. Still, this is how I generally try to do it.

First, consider the beta sheets. These are the "flat" pieces. A good fold tends to see many of them aligned with one another and sharing one or more hydrogen bonds. I've read that beta sheets sharing hydrogen bonds tend to have a higher bond strength (and thus presumably higher scores) when the sheets are on the inside of the protein among the hydrophobic bits, but I have not found this to be the case in fold.it proteins so far -- more often, the beta sheets seem to be on the outside. Even when I have put them on the inside on purpose, they seem to migrate outward as I tweak things for score increases.

You can't just h-bond the betas however you like, though. Above all, try to arrange them so that their hydrophilic stuff is more outside than in. Note that the easiest way to manipulate the shape of a beta is simply to grab the things it's attached to, and pull gently. Pulling the beta itself doesn't work well, because -- apparently -- there isn't the leverage given the tools we now have. The shape of a beta depends almost entirely on the configuration of things attached to and around it.

Strive for an aesthetically pleasing shape (I'm serious about this). Beta sheets flow. They're not like Le Corbusier's "efficient" buildings, but more like Frank Gehry's twisty and curvaceous architecture. Don't even try ironing them out. Once your beta sheets have an appropriately pleasing look, you'll want to try to fold everything else into a compact mass around them, with the hydrophobic elements facing inward. Do this slowly and gradually, and keep your hydrogen bonds intact by rubber banding them together while you work on other parts, otherwise the h-bonds will fall apart.

If you're totally mystified about beta sheet aesthetics, take a look at some proteins whose structures are already known. Ubiquitin is a work of art, as is insulin. Browse wikipedia for others. Rosetta@home is another excellent source of inspiration.

After you've dealt with beta sheet aesthetics and reached a decently compact form for everything else, shake the side chains. I may be wrong here, but I have found that shaking is not terribly useful, EXCEPT in two cases: after you've radically rejiggered the entire structure, or after you've improved your overall score. After little adjustments that either decrease your score or leave it the same, you almost never gain anything from shaking. Don't waste your time on it.

If you've got a good build, you will know it when you shake -- the score will go up a lot, sometimes by 100 points, sometimes even more. If this happens, try wiggling again. Sometimes the score will increase yet again, but there tend to be diminishing returns here. Once your protein reaches this point, reassess the aesthetics, the hydropilic/hydrophobic areas, and the compactness. Adjust as needed. Repeat until you get stuck. Then move to the next step...

Turn on the "show voids" option. Look at where the voids are, look at where you've got stuff dangling out, and try to make the one meet the other. Or else try to push the same element that's dangling in such a way that another bit will pop into a void. Whenever you achieve this, your score will go up. Be sure to shake the sidechains after every score gain you earn from this and other methods.

I seldom tweak the sidechains, because I find that 1) I don't gain much from it and 2) I can't easily predict what a tweak-wiggle combo will do. Indeed, it's pretty much a shot in the dark. Others disagree with me on this and tweak-wiggle as a way of moving the backbone elements. Me, if I want to move the backbone, left-click and drag it where I want it.

If you become dissatisfied with the fold you've created, do one of two things: Either work on a different puzzle, or go back to the beginning and try a different configuration. Either way, save your work, because you never know when an idea will occur to you about a structure that formerly looked unpromising.

Learning the interface is important in accomplishing all of these tasks. Here is what to do if you consistently have trouble making the protein do what you want it to do, even when you're not working in close quarters: Disregard the score and set yourself a goal for the protein's appearance. Use the tools to make a smiley face, or a letter, or a geometric form. Doing this for a while will teach you better how to use the interface to manipulate a protein, and then you'll be ready for the harder stuff.

Joined: 05/17/2008
Groups: None
Thanks for sharing your

Thanks for sharing your knowledge and experience, I'm sure it will be very helpful. It's cool that this project can be cooperative as well as competetive. Happy folding.

Joined: 05/20/2008
Groups: None
Show Voids? What's that?

Show Voids? What's that? Also, the demos didn't make it clear what parts are hydrophobic and what parts are hydrophilic. I'm doing pretty well given these holes in my knowledge, but I suspect I'd do better if I knew about them.

Diderot's picture
User offline. Last seen 6 years 43 weeks ago. Offline
Joined: 05/09/2008
Voids/Hydro

You can turn on "show voids" in the view menu. It will show you where the gaps are in your fold.

Hydrophobic sidechains can be made to glow orange, using the same menu. In fact, I think they may default to that. Hydrophilic ones will likewise glow blue. As a general rule -- but with many exceptions -- you want the hydrophobic ones on the inside (Good hydrophobic hiding!) and the hydrophilic ones on the outside.

Sitemap

Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Amazon, Microsoft, Adobe, RosettaCommons