9 replies [Last post]
Joined: 06/21/2008
Groups: None

I downloaded foldit and did the tutorials, but I'm completely lost when it comes to actually working on a real protein. What should I be looking for when I'm trying to raise my score? And what's stopping people from just clicking "wiggle" to raise their scores? (Because you can get a lot of points by just letting wiggle do its thing for a while.) Thanks. (And it's fun, when I know what I'm doing lol, and it IS for a good cause. I would, though, like to know what I'm looking for. =])

Joined: 05/11/2008
Groups: CASP8 team
several things to aim

several things to aim for:

1) minimize energy by making protein more compact
2) maximize h-bonds
3) wiggle and shake like mad =)

Joined: 05/11/2008
Groups: CASP8 team
several things to aim for:

several things to aim for:

1) minimize energy by making protein more compact
2) maximize h-bonds
3) wiggle and shake like mad =)

Joined: 06/21/2008
Groups: None
Thanks, I have another question.

Thanks. I have another question, though. How can you make hydrophobic side chains be on the inside and hydrophilic side chains be on the outside? I mean, I see how you can move parts of the backbone to get them to point more inwards/outwards, but you can't actually rotate parts of the backbone to put them inside. Do you have to just move the backbone to hide them then?

One more. :) Is there any specific strategy that anyone uses? I'm pretty much just guessing and shaking/wiggling a lot, without actually having a strategy. Anyone out there have a good strategy that a protein-newbie like me would find helpful? (After all, this is my first time benefiting all of humanity. =]) Thanks a lot.

Diderot's picture
User offline. Last seen 1 year 18 weeks ago. Offline
Joined: 05/09/2008
Wiggle, other strategies.

And what's stopping people from just clicking "wiggle" to raise their scores? (Because you can get a lot of points by just letting wiggle do its thing for a while.)

Wiggle helps a lot, and you should try it to get started. But... you will need to do many other things (rubber bands, nudges, rebuilds, locking, etc) to get into the higher scores.

Wiggle helps you understand where things "want" to go. But often they need to be guided so that the protein makes a harmonious whole. Look at some solved proteins and you'll get a sense of the aesthetics of it. Then see where things want to go using wiggle, and guide them into a good combination of their own inclinations and a compact, well-bonded, hydrophobic-hidden, aesthetically pleasing shape.

Angus's picture
User offline. Last seen 3 days 4 hours ago. Offline
Joined: 06/04/2008
Groups: Beta Folders
Solved proteins ?

Look at some solved proteins and you'll get a sense of the aesthetics of it.

How is this done? I haven't found a way to look at the "winning" solution for closed puzzles.

Diderot's picture
User offline. Last seen 1 year 18 weeks ago. Offline
Joined: 05/09/2008
Wikipedia.

Wikipedia.

feet1st's picture
User offline. Last seen 9 years 19 weeks ago. Offline
Joined: 01/25/2008
Groups: None
Protein Data Bank

To get a feel for what native protein structures look like, you can examine those recorded in the protein data bank.

Unfortunately, using that site is rather cryptic for the non-biotechie. You need a protein model viewer, and then to download the .pdb file of a protein of interest to you.

Perhaps someone who knows more about it could help us use the search facilities and etc. to file a protein in the database that is similar to any given protein we might be working on.

Angus's picture
User offline. Last seen 3 days 4 hours ago. Offline
Joined: 06/04/2008
Groups: Beta Folders
I was actually asking about

I was actually asking about how to view - in the application so it can be rotated around - the best "winning" protein from the ones that we've worked on here.

feet1st's picture
User offline. Last seen 9 years 19 weeks ago. Offline
Joined: 01/25/2008
Groups: None
I've suggested that

I have suggested previously they allow import of the best solutions, and haven't seen it available yet.

But showing the images, like they did in the news post... apparently just long enough ago it is no longer on the homepage.

Another step in this direction was the addition of the "load as model" feature, which allows you to load your protein, and load another as a frame of reference.

Another place where you can see native protein structures (on some tasks) is running Rosetta@home. You can enlarge and rotate the native structure, when it is known. So, in addition to wanting to see what the best score looks like, I feel there is much to be learned by careful study of a library of native protein structures.

I read elsewhere that they are working on actually computing the RMSD of the models. This is a measure of how good a given model matches the native structure. The "score" in the game, and the energy function behind it, is all you have to gauge by when the protein's structure is unknown. But once the structure is solved by X-ray or NMR, the RMSD gives you a better idea if your model might be "close enough" to be useful in making predictions for potential drug development. RMSD below about 1.5 is a very good fit. And a RMSD of zero is a perfect match.

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, Boehringer Ingelheim, RosettaCommons