Presentation of raw ED data and ED tools

Case number:845813-993838
Topic:Game: Tools
Opened by:Timo van der Laan
Status:Open
Type:Suggestion
Opened on:Thursday, November 1, 2012 - 14:51
Last modified:Friday, March 8, 2013 - 14:51

In the previous ED puzzles I was able to use the ED chart as a guideline.
With 647 I find this impossible to do.
Maybe some extra tooling would help.
First of all adding the 3 translation vectors in the picture might help a little to find the main orientation.
And second adding tooling to make it possible to manipulate the visual representation.
Blanking out parts or colouring parts by moving a box around the part you want to blank out or colour. That way you can try to make the picture more usable as a guideline.

(Thu, 11/01/2012 - 14:51  |  13 comments)


Susume's picture
User offline. Last seen 3 days 11 hours ago. Offline
Joined: 10/02/2011

It is difficult to tell where to "wrap" the protein into the adjacent cell, since we can't see the ED cloud extending past the viewing prism on that side. Would it be possible to scroll through the cloud, or drag a prism frame across the cloud, so we can see the extension of the protein into the adjacent cell? The amount of cloud showing at any one instant could be the same, but the cloud could disappear on the left and more cloud appear from the right as we slide the frame in that direction. Since the game knows where the continuation of the cloud is, let us see different subsets of it depending which piece of the protein we are working on.

Joined: 09/21/2011
Groups: Void Crushers

Agree, this, together with a blank out tool would make it possible to create a more or less coherent view. Or give us a cell of size 2*2*2 with a blank out tool.

jflat06's picture
User offline. Last seen 21 hours 32 min ago. Offline
Joined: 09/29/2010
Groups: Window Group

I definitely agree there need to be more options for adjusting the view of the density. I'm looking into implementing some of these.

Susume's picture
User offline. Last seen 3 days 11 hours ago. Offline
Joined: 10/02/2011

The native pdb (4GZV) for puzzle 647 shows a dimer of tetramers. Which begs the question, does the ED we were given show a single protein tiled in space, or a tetramer tiled in space? Will we know, in general, whether we are looking at repeats of a single protein?

Joined: 09/21/2011
Groups: Void Crushers

Now I understand how the cloud is much bigger as the protein. The cload shows a dimer.... Making it really impossible for us to get near.

Susume's picture
User offline. Last seen 3 days 11 hours ago. Offline
Joined: 10/02/2011

For a puzzle like 663 it becomes really important to have a blank out tool so we can narrow down the portion of the ED we are trying to match. Timo's ideas of 1) showing the three axes and 2) letting us color portions of the ED would help greatly too. Coloring could be used to mark features we have found, as well as to mark the path of the backbone in an area before trying to match it with a piece of the protein.

Susume's picture
User offline. Last seen 3 days 11 hours ago. Offline
Joined: 10/02/2011

I hope more people will weigh in with their wishes for using the ED map. I find over and over that I need to mark an area of the map so that I can zoom out, rotate things around, find an appropriate section of protein, cut it out and move it into the marked location. I also need to mark an area so I can identify it again after real world interruptions, since the game has to take a backseat to other tasks.

You could kill two birds with one stone by allowing us to select a portion of the cloud and set both the color and alpha for that portion. For areas we wish to hide, we could set alpha (transparency) to full. For other areas we could come up with our own conventions of marking landmarks using different colors. A button to reset alpha to default for the whole map (without resetting the colors) would let us find the hidden portions again. For completeness, another button could reset all colors to one value. An undo button for map marking (as opposed to protein changes) would also help.

Joined: 09/18/2009
Groups: SETI.Germany

@Susume:
A kind of user-defined palette?
Assgin each measured density value to a user-defined RGBa- or HSLa-color value?

Susume's picture
User offline. Last seen 3 days 11 hours ago. Offline
Joined: 10/02/2011

I was not thinking of using color to mark density values or any feature the game could identify for us. Rather, let us use color (or something else) to manually mark areas of the cloud that we identify by examination. Personally I would want one color for helices I had found, one for sheets and one for large hydrophobes. Different people might want to mark different landmarks.

brow42's picture
User offline. Last seen 3 weeks 2 days ago. Offline
Joined: 09/19/2011
Groups: None

It's fine to ask for features but you also have to think of a user interface for the feature, and hopefully not adding more input methods or check boxes. It's hard to think of an easy way to paint parts of the ED.

Maybe we could have a 'spherical' brush, like the circular airbrush (dense in the middle, fades at the edges) in Photoshop. You'd still be kinda stuck with 2-D painting, with an awkward depth control with shift click or something. With practice, people might be able to rotate the ED so the part they want to paint is in the plane of the screen, but it would still be annoying to player to paint-rotate-paint-rotate.

Susume's picture
User offline. Last seen 3 days 11 hours ago. Offline
Joined: 10/02/2011

One feature I learned from a teammate definitely helps with visibility - hover mouse over a segment and press shift-Q to get a cutaway view of that part of the protein and the cloud around it. Q returns to full view. This info has been so helpful that maybe it should be mentioned in puzzle comments for ED puzzles.

To set the center of rotation in the cloud, place a segment in the cloud, hover over it with mouse and press Q.

As a workaround for being unable to mark the cloud directly, you can band from a nearby segment to a feature in the cloud you want to mark, and add a note to the segment you banded from.

Hanto's picture
User offline. Last seen 4 weeks 1 day ago. Offline
Joined: 05/10/2008
Groups: None

Hovering mouse over segment and pressing shift-Q only displaces the puzzle off-center in my case and continuing to hold the shift-Q only forces the puzzle to scroll forward until the puzzle reaches full screen and beyond of my client. In other words I am not getting ANY indication of a cutaway view of any part of the puzzle. I have to assume that in this case ' cutaway ' may mean enlargement of the area under the hovering mouse. And I tend to agree with Timo and Brow, in the context that these clouds may be showing information beyond the protein itself and as such might be adding a great deal of confusion to the process of generating solutions that represent good scores. The few mediocre score that I have been able to create were created by means I would not brag about and would never work on other types of puzzles.
These ED puzzles seem to be more van Gogh then science and as such, I intend to stop playing them until they do become science with human eyes that can perceive what is happening and/or tools that can make sense of what van Gogh is trying to tell us.

Joined: 09/24/2012
Groups: Go Science

Personally, as a beginner with ED, I do not understand neither "see" what to do with this electron cloud. When pushing on "align to ED", I loose points, then I get small points by clicking and clinking again 1000 times on "refine alignment" button. Then I give up and I simply run recipes.
Some guideline on what to do with this electron cloud would be usefull. Anything that would clearly show where our protein should match (and allow us to make something "human") would be welcome !

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