Request: Structure Initialization Tool

Case number:845813-997105
Topic:Game: Tools
Opened by:brow42
Opened on:Monday, February 24, 2014 - 20:35
Last modified:Tuesday, March 4, 2014 - 21:14

Design is very difficult for both new and experienced players. Remember when Foldit was set up for the gamification of the folding problem. We had shake, wiggle, and a structure painter. We have a mutation wheel but also the M command can take care of that for us.

But design is more difficult. Or rather, we can design neat structures and make them score high in rosetta, but actual STABLE designs requires too much player knowledge. You've tried to guide us with the filters. But now we have the problem where we have to set the primary sequence for rebuild to make a helix. Before, if we could get a helix, the mutate tool would then find an appropriate sequence.

I'm sure the response to this will be to tweak the fragment library again. But that's the wrong solution.

How to make a helix (according to foldit tutorials):

1) Paint the helix structure in structure mode
2) Rebuild
3) There is no 3, because you're done.

This is what needs to happen, especially in design mode. All this stuff about MALAK and fragment libraries is technical stuff that new players don't necessarily know about, and shouldn't need to know about. If a player wants a helix, they should get a helix. Let rosetta worry about the stability.

Here's how the proposed tool should work. It can basically be a part of the rebuild tool, as it preserves most of the original behavior:

1) If the rebuild selection does not have a free end, rebuild as normal.
2) If the rebuild selection is mixed SS, rebuild as normal.
3) If the rebuild selection is all one SS, and is flanked by loops, and one of those loops has a free end (a cutpoint or terminal) then just set all the non-loop residues to their ideal angles (loop angles remain unchanged). Just set them. Not 3 at a time, no fragment library, no cis bonds. Just a helix/sheet. Prolines will clash, fine.

At this point the player can place the helix/sheet and ask the mutate tool to find a reasonable sequence, and work on healing the cut point. Maybe they succeed, maybe not, but a huge amount of frustration evaporates.

(Mon, 02/24/2014 - 20:35  |  12 comments)

Joined: 04/15/2012
Groups: Beta Folders

Seems rather similar to this one... ;)

Joined: 09/21/2011

yes. first let me do what i intend to do quickly and easily, then the game can respond to it however its designed to. if i intend to make a helix/sheet, i don't want to spend all day holding down a tweak button, and rebuild is like cranking a jack-in-the-box and hoping it pops up like you want it to.
gaming efficiency is good for the science end too. you find out quickly what a terrible job we did on the protein ;). plus we can start over more often when it is a quicker build. playability is all about how easily we can interact with the gaming environment. when newbs come in to play and they say "this is hard", they're not talking about the science, they're talking about the difficulty in executing even the simplest of tasks.
the cut tool is a perfect example. what a great way to make proteins easier to manipulate. it allowed us freedom to manipulate the protein quickly and easily.
if you give us the very best strategies in the world for brain surgery, but hand us an oven mitt and a spoon...

spmm's picture
User offline. Last seen 18 weeks 3 days ago. Offline
Joined: 08/05/2010
Groups: Void Crushers

In the olden days, in main interface only, if two helices had a three seg loop you could double click the loop and all three sections would rebuild. You can still do this but the helices never find "snuggle" positions anymore. This was different to selecting the three portions and rebuilding which doesn't snuggle.

A recognisable loop shape emerged and the helices sat really well together, I still don't really get 'handedness' in sets of helices but it may have something to do with that. There were usually two extra good positions.
I still hop into main to try it but it is rare to see that loop pose emerge and a lot of work remains to do on the helices. Cuts in those loops never really come good, but that is probably the wetware. This is not new in NC.

jflat06's picture
User offline. Last seen 6 hours 3 min ago. Offline
Joined: 09/29/2010
Groups: Window Group

I agree that it should be easier to do this. Our tutorial about using Rebuild to make secondary structures is misleading because that's really not what Rebuild was designed to do.

Rebuild inserts 3-res fragments into random subsets of the sequence, and for longer sequences, this can take ages. You're basically just sitting there hoping that Rebuild picks the correct part of the sequence to do an insertion on (and picking the right fragment, and hoping the new structure scores better in terms of the energy evaluation).

Tweak is the better option, and I'm working on fixing Tweak to form secondary structures much faster and fixing some bugs which produce strange artifacts. I'm also working on some other options which might be better.

Joined: 04/15/2012
Groups: Beta Folders

Honestly, I would rather just get my structure and be done with it-I honestly have NO clue how to use tweak to do anything aside from twisting a structure around and doing a register shift.

Joined: 04/15/2012
Groups: Beta Folders

(Which may be of my own fault, but I haven't seen that explained, for something that would be rather critical...)

Susume's picture
User offline. Last seen 4 hours 38 min ago. Offline
Joined: 10/02/2011

There is a purple ball on the tweak tool (which is not shown in the tutorial) that straightens the sheet or helix you are working on. This can be used to straighten even an extremely bad helix (or an extended chain) into a perfect coil, if you hold it down long enough, assuming the ends are free to move.

Maybe we need a new tutorial on how to make a straight helix, and one on how to straighten a sheet.

Joined: 04/15/2012
Groups: Beta Folders

Oh! I had no clue... Thanks for letting me know that, this would've been nice to know.

And re-making note of my thought:

Would ease the change into full game, and make for cleaner and less intimidating learning.

jflat06's picture
User offline. Last seen 6 hours 3 min ago. Offline
Joined: 09/29/2010
Groups: Window Group

Yeah, this is one of the biggest problems right now. We tell people to use Rebuild in the tutorials, and nowhere does it really tell you about this feature of tweak. It's also still awkward to use tweak for this, so I'm looking at making it easier.

Joined: 09/21/2011

an important fix to keep in mind if you're looking in to this.
its not just how long tweak takes. if there are other structures on the protein, tweak can't complete its mission. the other structures stay in a fixed location and hold the protein in place. you can't really use the tweak tool unless there is an abundance of slack in the loops that connect to the tweaked structure (assuming those loops don't get all twisted and knotted).
i see no reason why the whole protein can't move along with the tweak. if all the other structures are converted to loop then it will perform this way (clumsily).
you can imagine how self defeating it is to remove all your other structures so that you can tweak just one structure.
this fix may be more important than the speed of the tweak, and it might be easier to implement.

Joined: 04/15/2012
Groups: Beta Folders

If so, I'd want to be able to turn it on and off...

Could you use cutpoints to do this?

Joined: 09/21/2011

yes cutpoints, or freeze seg at point where you want to lock it down.


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