What makes you noticed?

Case number:699969-992444
Opened by:AsDawnBreaks
Opened on:Friday, April 20, 2012 - 22:26
Last modified:Sunday, October 6, 2013 - 16:08

What makes your protein noticed and used in real-life purpices? Is it the final protein the highest scoring user has? How often does it happen? What can I do to increase my chance of being helpful?

(Fri, 04/20/2012 - 22:26  |  10 comments)

marie_s's picture
User offline. Last seen 1 year 28 weeks ago. Offline
Joined: 05/18/2008
Groups: None

Is it the final protein the highest scoring user has?
no see here for an example http://foldit.wikia.com/wiki/CASP_9

What can I do to increase my chance of being helpful?
you can see the wiki for other player strategies
but dont forget you have to find your own strategy to find a solution different for any other players and from Rosetta

you can also be helpful in writing scripts with new ideas (see other scripts before or ask in global f your idea is original), share your result at the end of each puzzle when your score is high, ...

Joined: 04/15/2012
Groups: Beta Folders

It says: Baker Lab will choose the best solutions for each puzzle and submit them, where they will be placed on a server with other groups solutions.

What server, and how are you picked? It soundes like the highest scored to me.

Susume's picture
User offline. Last seen 1 day 23 hours ago. Offline
Joined: 10/02/2011

It is not just the highest scoring solutions that are submitted. The scientists are looking for solutions that differ noticeably from the server model they started from. You can read more about it in beta_helix' long post here: http://fold.it/portal/node/991921

On one recent CASP puzzle the scientists went as low as rank 75 to submit a solution that differed appreciably from the server models, though more often they pick things in the top 20 or so ranks. My sense is that they don't just want all the loops to be moved a little bit (which is what typically happens if you use all scripts and no hand folding), but to see one or more major structures (sheets, helices, long loops) moved to entirely new positions or swapped with other structures. In general so far, it seems they are picking two solutions from the Rosetta prediction puzzle and three from the Quark & Zhang prediction puzzle for each CASP Roll target. So it is worth playing both puzzles, and moving pieces of the backbone around, even if you don't rank in the top ten.

The place they send the solutions is predictioncenter.org, where you can read about the CASP competitions and see predictions and results from prior years' competitions. You can also download other server predictions for current CASP Roll puzzles, but you would need a protein viewer program such as pymol to look at them. We probably won't get a look at the other human teams' results until December when the conference to discuss the results is held.

Joined: 04/15/2012
Groups: Beta Folders

So they just look at the last protein or best protein folded by the user befor the challenge closes.

infjamc's picture
User offline. Last seen 5 hours 22 min ago. Offline
Joined: 02/20/2009
Groups: Contenders

For another example where the solution that looked at is not the highest-scoring one, consider this:

This is the monkey virus protein that made the headlines last September. In this case, the solution that allowed the researchers to solve for the crystal structure came from the third-placed team.

jflat06's picture
User offline. Last seen 2 days 8 hours ago. Offline
Joined: 09/29/2010
Groups: Window Group

This is a very good set of questions. I'm not completely qualified to answer in detail, but I can say that our analysis depends on the particular puzzle we're running.

For prediction puzzles where we know the answer, and are attempting to refine our techniques, we generally create a plot such as the one attached. This plot shows a green dot for each solution committed to our server. The Y axis is the rosetta energy (lower is better), and the X axis is how "close" that solution is to the native (the native is shown as the blue blob. The further right you are, the better). As you can see, in this particular puzzle, players didn't get very close, so we take that into consideration, and attempt to make changes for the next iteration.


For prediction puzzles where we don't know the answer (such as the monkey virus protein), we look at the top solutions for each player/group, and also do some clustering of similar structures, and look for clusters that score well. We review these solutions both manually (to some extent) and using server methods, looking for promising candidates.

For design puzzles, much of the analysis is done by the scientists who were working on the particular puzzle. They go through the top solutions for each player/group, and filter these for promising solutions using their expertise. As has been stated above, it isn't all about the score. More qualitative things like secondary structure, hydrogen bonding, and packing are signs of a good protein. If you're doing well in the qualitative sense, chances are that your solution is good, even if you haven't quite managed to crank the points out of it.

Joined: 04/15/2012
Groups: Beta Folders

That's great. What is meandt by rosetta energy?

B_2's picture
User offline. Last seen 6 years 26 weeks ago. Offline
Joined: 11/29/2008
Groups: None
Joined: 04/15/2012
Groups: Beta Folders

Nice. Thanks!

Joined: 04/15/2012
Groups: Beta Folders
Status: Open » Closed

Closed, old, solved.


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