Align Guide mode

Case number:845813-990782
Topic:Game: Tools
Opened by:tristanlbailey
Status:Open
Type:Suggestion
Opened on:Sunday, October 9, 2011 - 14:27
Last modified:Monday, June 3, 2013 - 08:02

The current general Align guide tool (for the entire protein) rarely appears to help you out. Its current implementation to automatically re-align the guide when you click on it, is poor. Unless approximately 90% of the protein structure has already been manually aligned to the align guide (sometimes through hours of hard work), it will NOT correctly align to the majority of your protein structure when clicked. Since the Undo mechanism is apparently NOT related to the align tool, you cannot reset the align guide back to where it was, before the Align Guide button is clicked.

The local version of the tool is also not very helpful. When the user right-clicks on a particular segment to use the Align Guide tool relational to that segment, if the tool decides to work at all (sometimes it doesn't), it appears to align the guide in a completely different direction to where the majority of the protein structure lies at the time.

Apart from the recommendation to fix the Align Guide tool so that it is more user intuitive, I am going to suggest an Align Guide mode be implemented. Accessed through the Modes tab, the mode will allow the user to rotate and translocate the guide for the protein, to wherever he/she wishes. While some users will opt to have the guide sit directly on top of the majority of the protein structure, others may take a different approach, re-positioning the guide to the side of the protein structure as if it were a reverse mirror image, or silhouette of the structure.

An extension to this suggestion, is a Band Snap to Guide option, which could be enabled and disabled by the user, while using the Align Guide mode. Bands for each individual segment of a protein will automatically snap to the equivalent section on the guide, making it far easier to align the structure to the guide.

Please note that the alignment of the protein structure to the guide, and the alignment of the guide to the protein structure, are effectively two separate, yet equally important concepts presented here.

(Sun, 10/09/2011 - 14:27  |  15 comments)


dedfed's picture
User offline. Last seen 7 years 21 weeks ago. Offline
Joined: 05/03/2012
Groups: None

i was actually just thinking about this, had to go back and see if anyone had talked about it, i hadn't thought about possible solutions, it's just, a bit ago i was trying to align my protein to the guide as best i could, and after a while of tweaking and adjusting, it looked perfect, i went to freeze the section, and let go a little bit to the right, accidentally selecting align guide, so the whole thing shifted off in another direction. so of course, i did the usual thing, and clicked undo, and it undid my last move. the guide stayed in its place though. it would be very beneficial to be able to move the guide at times, get it where i want to start it with, and then start adjusting the protein around it. at least an undo last mistake would be nice :)

brow42's picture
User offline. Last seen 6 hours 29 min ago. Offline
Joined: 09/19/2011
Groups: None

An extension to this suggestion, is a Band Snap to Guide option, which could be enabled and disabled by the user, while using the Align Guide mode. Bands for each individual segment of a protein will automatically snap to the equivalent section on the guide, making it far easier to align the structure to the guide.

Doesn't this turn a guide puzzle into an almost fully automated process? Banding is annoying and a lot of work but that's sort of the entire puzzle!

Agree about the uselessness of the align guide, though.

spmm's picture
User offline. Last seen 40 weeks 4 days ago. Offline
Joined: 08/05/2010
Groups: Void Crushers

Some new players, myself included, would like to have the guide set to the side so that they can see the protein and the guide separately, I also really wanted to just grab it with both hands to manipulate it; I understand work is in progress with Kinect to realise this:)

As offsetting the guide wasn't possible, I just pulled the protein away from the guide and took two screen shots so that I could look at the sheet configurations, then reset best and used the guide in situ and the reference shots.
In the next puzzle I had learnt to use the side chains to reference the guide and begun to understand a few better ways to move the protein around, I didn't need to take the screen shots to show the full guide 'picture'. Note that I was not using scripts at that time, just working manually.

The Align guide works fine, if you click a segment the guide will align to that section which can be very useful, as you can align each section, even loop sections to match them perfectly.

Yes, align guide does not respond to undo - however most people do not want the guide to move once they have started moving towards it.

A band script to band the protein to each segment of the guide? Why bother doing the puzzle in the first place?
Perhaps we are discussing two things:
1 - learning new skills as a new player, who perhaps like me doesn't really have much of an idea of what a good protein looks like, or how to move things around very well.
or
2 - copying existing solos or evos for those in groups - certainly a 'band exactly to high scoring solo' would be handy for those who chose to use it.

QTTN is the next step from the tutorials, a shape to aim towards given the tools you have just learnt.

spmm's picture
User offline. Last seen 40 weeks 4 days ago. Offline
Joined: 08/05/2010
Groups: Void Crushers

''Unless approximately 90% of the protein structure has already been manually aligned to the align guide (sometimes through hours of hard work), it will NOT correctly align to the majority of your protein structure when clicked.''

Exactly - the guide is a fixed configuration, the native - the aim of the puzzle is for you to match the guide, you will score well if you can do that. And that means that every sidechain, every part of every section is aligned. Once you have done that you can explore and possibly gain more points.

Hours of hard work? took me days, but that was/is the fun part for me.

You may of course do something different which is great, especially if you can get it to score, you can do the QTTN puzzles with the guide turned off and still score well, your choice.

Joined: 09/22/2011
Groups: None

Wow, this post went back a while ago...
When I first started playing Foldit, it wasn't too long before I made this feedback, so it was essentially one of my first feedback posts. I was obviously still learning the ropes back then, and this was probably something that frustrated me considerably to want to do that, hence the reason why I avoid QTTN puzzles nowadays.
The point of the matter is that although I understand there are people that like these puzzles (most likely the experienced players), other players like myself became quickly turned off by the QTTN puzzles, and the align guide along with it. Like most other players, I have fun from getting a fairly high rank, but also have fun and interest when I know I could be helping cure disease, and the like. From this perspective, I don't enjoy QTTN puzzles at all.
I suspect that these puzzles are supposedly meant for fun only, nothing more.
If I were to take the perspective that the developers were somehow gleaning information about how people fold from these puzzles, I would not understand, because we are working with known (or made up) structures. If perhaps they wanted to gauge how quickly players rage-quit from these kinds puzzles, while working with clunky tools, then I would understand.
If there is, in fact another objective to solving these puzzles (apart from trying to match the guide as best as possible, and scoring high), then it should be made clearer.
Spmm, you raised a good point; if you can still score well without following the guide, then why bother following it at all? Sure, it might speed up the path to a good score to work with it initially, but when you are trying to make the top 10, or even the top 20, it's no longer a matter of how well you can closely follow the guide, but instead how best to continue folding without it. If I am right about saying this, then this in itself should be made clearer to beginners, so they don't tear hairs out of their heads wondering why their seemingly perfectly aligned protein that they spent hours on, is not scoring as highly as they would have liked.
Days? You're crazy. :P

Joined: 04/28/2013
Groups: None

I like the term rage-quit. Never heard it before. As a newbie I find the tools leave a lot to be desired. As an educator, I think that spending some more time focusing on the tools - especially fixing their clunkiness, the fact that they don't go away easily, they repeat, and they don't give enough information - would be helpful. It would certainly raise foldit's whuffie in in the Out There...this tool has promise. Don't neglect continuing to refine the opening moves. The likelihood of your program making good solid contributions increases exponentially with the number of players playing and the depth to which they understand the logic behind what they're doing. Good tools in the beginning that WORK will go a long way toward that. As of May 19, 2013, they don't work well enough yet.

Joined: 04/15/2012
Groups: Beta Folders

The align guide does seem useless to me. You can really have the protein do whatever you want, so it is hard to get it in the correct position. I want part a to go left, and part b twists itself into a knot. Really tedious. It is ok if it takes work, but it has to be able to do what I want it to do without turning inside-out.

jflat06's picture
User offline. Last seen 17 hours 27 min ago. Offline
Joined: 09/29/2010
Groups: Window Group

Initially these puzzles were a means for us to verify that, given the native, the tools of Foldit were enough for players to reach the native state. Since then, they've been kept going both as easier puzzles for newer players, and also as a learning tool. It is important for people to be able to see ACTUAL structures, and observe patterns in how proteins fold natively.

As for the align guide tool, have you tried using the local alignment? If you align, say, a particular helix, you can get a much closer match to that helix, and you can see just where the loops on either side go awry.

A possibility that may help is if we enable the move tool on QTTNs, so that you can do manual alignment (or even move it off to the side, if that's what you want).

Joined: 04/15/2012
Groups: Beta Folders

What might help if there was an option for when you freeze, for it to stay in place. I know this happens when a segment is locked, but I don't think it is available for users to do. I want to see one section, fix it, keep it where it is, and move on doing the next without messing it up. Like I said before, you move one part, and it knots another part.

infjamc's picture
User offline. Last seen 2 years 45 weeks ago. Offline
Joined: 02/20/2009
Groups: Contenders

There is a workaround: Bands of strength 10.

1. First, add a band to the segment and make the length of the band as short as possible.
2. Increase the band strength to 10 via a recipe or script.
3. Result: when the band is enabled, movement for the segment attached to it will be restricted.

If you have the patience to do so, you can even match the guide exactly by putting such bands on every amino acid and wiggling (note that you'll have to do the same for side chains as well, as it's possible for the backbone to be at the correct location while the side chain is pointing in the opposite direction of what the guide indicates).

spmm's picture
User offline. Last seen 40 weeks 4 days ago. Offline
Joined: 08/05/2010
Groups: Void Crushers

thanks jflat - I think the move tool would help so that the guide can be fully exposed, to better understand the pose and to adjust the secondary structure if necessary.

Joined: 04/15/2012
Groups: Beta Folders

Good suggestions. My only other issue then is that you see to much of the puzzle and the protein all at once. Is there a way that it could be made so that you can only see one part of the protein at a time? Maybe a hide feature, where when you hide a part, it locks, and it turns almost clear, exept for a very vague outline. I think it would help if you want to work on one section without any others in the way.

infjamc's picture
User offline. Last seen 2 years 45 weeks ago. Offline
Joined: 02/20/2009
Groups: Contenders

Move the mouse over a segment, then press Shift-Q. This would hide everything except for the atoms very close to it. To restore the normal viewing mode, press Q anywhere.

Joined: 04/15/2012
Groups: Beta Folders

Perfect! Thanks!!!

Joined: 09/21/2011
Groups: Void Crushers

The most frustrating thing with aligning to the guide for me is that the A key in selection interface moves the guide ruining the position of the bands without a possibility to undo.
(Without a guide A gives the highest scoring position)

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