6 replies [Last post]
Susume's picture
User offline. Last seen 5 hours 27 min ago. Offline
Joined: 10/02/2011

In the Nature paper, the authors list foldit's Rebuild among tools for manipulating secondary structure, along with Tweak. This doesn't match how I use Rebuild, so I thought I would pose the question here - how do you usually use the Rebuild tool?

I occasionally rebuild an area by hand (although now I prefer remix for hand work), but by far the most use I make of rebuild is in scripts, primarily TvdL enhanced DRW but also some other rebuilder scripts. These scripts function to randomly (more or less) sample a large number of backbone shapes near the current shape. I crank the number of rebuilds per area up to 45 on DRW (I modified the slider to allow this), because I found way back during NewChapter testing that the more rebuilds you try on an area the better the chance of finding points. During the course of a puzzle, regardless of puzzle type, I run enough DRW to generate and compare many thousands of nearby backbone positions, covering the entire length of the protein multiple times. The main reason the old hand-fold-only puzzles were so tedious is that we had to do this sampling by hand, and one's score was largely a product of how many hours one was willing to spend rebuilding the whole protein area by area, by hand, to sample nearby backbone shapes. If the developers want to wean us off rebuild, the best way is to provide another tool that allows this kind of large-scale sampling. IMO the remix library of backbone positions is not large enough yet to compete.

During early game I leave the SS in place while running DRW, but in mid and late game I run it with the default setting changing the whole protein to loop first, so it is not even using SS. Do you use rebuild primarily to manipulate secondary structure?

KarenCH's picture
User offline. Last seen 31 weeks 12 hours ago. Offline
Joined: 07/14/2012
Mostly I do things similarly

Mostly I do things similarly - I leave SS in place when I run DRW or other rebuild-based scripts. It scores better and faster that way.

bertro's picture
User offline. Last seen 10 hours 9 min ago. Offline
Joined: 05/02/2011
Groups: Beta Folders
Also this
Joined: 10/30/2012
Groups: Beta Folders
other than scripting, edrw

other than scripting, edrw and helix rebuild, mostly use it for closing cuts, select area I want close, then s/w, rebuild, close and then remix

frood66's picture
User offline. Last seen 7 hours 22 min ago. Offline
Joined: 09/20/2011
Groups: Marvin's bunch
I take a very similar approach to Subsume

increasing the number of rebuilds does not slow things by much. I have commented B4 that the remix route appears to be a better concept - though the library so far seems to be a little small. I might also add that I find 'manual rebuild' (thx Brow) very useful in early work.

But I have one thought - if rebuild were to disappear (not saying it will) and we depended on a remix based rework - would this result in less original folds? Just a thought.

Joined: 06/06/2013
Groups: Gargleblasters
rebuild

I will spot rebuild some bad sections if I have time. If not, lots of DRW. Andlike others, I keep SS early on and let it go to loop only towards the end.
DRW allows me to keep a shape early game to see if it works. however, with a slow machine, I can't keep up with those of you running tons of rebuilds on various structures. So if I guess wrong by the end of about 24 hours, the game is over for me. So does this become a proxy for computer power? ok that it does. just a question. after all, hand fold will not change dramatically, though score might differ quite a bit

Joined: 09/24/2012
Groups: Go Science
seldom manual rebuild

I use remix manually to try to correct loops, selecting on the visual aspect, rama and filters, regardless of total score (after first design, I only keep a remix-proposed one if possible, then s/w).

Only if there is no remix solution, I may use manual rebuild (max about 15 times - no patience), selecting on the visual aspect, rama and filters, regardless of total score. Then s/w reversing on several backward rebuild solutions.

After the first hour till end, I mainly use recipes:
-Quaking remix the first days (or manual remix on middle-end game)
-some times eDRemix on loops only, seldom on all protein
-eDRW (default change to loop) unless on early game or when I want to keep a SS for any reason (e.g. SS or fragment filters, or when my SS is not the same as the predicted one).

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