Need more "human intuitive" protein manipulations

Case number:845799-985775
Topic:Biochem
Opened by:SSquires
Status:Open
Type:Suggestion
Opened on:Tuesday, March 10, 2009 - 15:48
Last modified:Wednesday, June 27, 2012 - 13:00

The description of the objective of "Fold It!" says:

"We’re collecting data to find out if humans' pattern-recognition and puzzle-solving abilities make them more efficient than existing computer programs at pattern-folding tasks."

After playing the game for a while, I'm convinced that the game features provided to manipulate proteins are not well matched or well suited to how humans think or changes that humans want to achieve in manipulating the proteins. Simply put, the mouse and keyboard tools provide a "hammer", when the user wants a "wrench".

Examples:

1. I've often wanted to "move the coil over here...", or even "move the coil to the other side of the sheet...", or "rearrange the order of the sheets...". That is, my thinking is of movement of gross structures (coils and sheets) in major ways. The primitive "freeze some things", "pull some other things with the mouse", and "repeat that many times" is extremely time consuming, and almost never works successfully.
* The molecule is often "too stiff", and won't allow starting motions I want to do. There should be a mode where the molecule is completely pliable (easy to make ANY motion), and then a mode that returns it to an "allowable physical configuration" in the selected limited area of the chain.
* The section of the chain I tug on can move in an unintended direction and in a jerky, out of control manner. This is probably part of the same issue in the prior bullet.

Suggestion: offer a mode where short sections of the chain can be made completely pliable with unrestricted movements, and other sections can be grouped and "frozen as a group, so they always move as a group, but are not frozen in their global position in space". I would make a separate view for this purpose. Also, I think that in this view, the individual amino acid component lengths should be allowed to stretch to any length during interim movements, with a means for the human operator to control the return of each/all sections to the standard length after the macro-movements of large parts of the protein have been accomplished. Grouping might have these "stretchable" portions of the chain as a default for un-frozen items, or might require explicit identification as a separate concept (vs. "frozen" and "normal stiffness"). Grouping of part of the chain should be able to be recalled and sectional grouping should be able to be turned on and off, without having to manually re-group from scratch. Grouping at a coil and sheet level should be a very simple process and require little time and effort from the user.

2. "Hydrophobics need to be on the inside" is an easily understood human concept, and is very important for game scoring, but trying to accomplish that with the limited mouse motions and keystrokes is a nightmare. The control to turn (rotate or flip) a coil or sheet almost never works... seemingly because the rest of the molecule is not pliable enough and will not compensate to accomodate the motion...

Suggestion: Either make the rotation/flipping/sliding control usable, or create something (probably 3 separate tools!) that does work easily.

3. Sometimes, movement of a part of the chain with a mouse will create conflicts that cause an 'explosion' when a subsequent global or local wiggle is started. While such motion is 'realistic' of what might happen in the real world when atoms try to occupy the same space at the same time, the explosions in the game are very non-productive, and not helpful. The recovery from an explosion is almost always an 'undo' or 'reset', and try the same thing again, hoping to guess when the mouse motion is 'just far enough to avoid explosive conflicts' ... an ill defined behaviour.

Suggestion: The game should have one or more standard, user selectable methods to resolve conflicts in a 'non-explosive' manner. One of these could have the computer guess a 'nearby' configuration that's non-explosive. Another might have a 'move atoms in 'exploding directions' (WITHOUT the 'energy dynamics' implied by the word explosion) until all molecular strain forces are below an 'explosive level'. etc.

4. Moving small sections of the chain that are not part of helices or sheets is difficult and tedious. The required multiple steps of freezing, unfreezing, and uncertain mouse movements - all result in a lot of wasted user time, that is totally unimportant to the Fold It project goal.

Suggestion: Offer something like: 1. a selection of alternative shapes in a library side-window, 2: a definition of interpolation points for a spline by mouse selection of points for the section, or 3. Computer generation of alternative configurations ONE AT A TIME - with the user able to stop at the selected one.

5. Humans intuitively understand that the solution to forcing compaction of atoms is to "SQUEEZE a bunch of the atoms together" by pressing simultaneously on opposides of the molecule.

Suggestion: Offer a tool that allows humans to SQUEEZE part of the molecule... without producing conflicts that result in explosive reactions. Probably the squeeze tools needs to have a pressure regulator, that defaults to a very low pressure level, but allows the human to increase the pressure in steps, and undo the last step, if the pressure produced too large a change or an undesired result...

(Tue, 03/10/2009 - 15:48  |  2 comments)


axcho's picture
User offline. Last seen 7 years 23 weeks ago. Offline
Joined: 12/18/2007
Groups: None

I think you bring up some very good points and I would very much like to see some experimentation with these suggestions. Even if it doesn't work and such experiments do not end up producing useful tools or interactions for folders, they would demonstrate why it doesn't work, which would be useful in itself.

spmm's picture
User offline. Last seen 40 weeks 3 days ago. Offline
Joined: 08/05/2010
Groups: Void Crushers

just thought this may be of interest: quotes from above text:
We’re collecting data to find out if humans' pattern-recognition and puzzle-solving abilities make them more efficient than existing computer programs at pattern-folding tasks."

Suggestion: offer a mode where short sections of the chain can be made completely pliable with unrestricted movements, and other sections can be grouped and "frozen as a group, so they always move as a group, but are not frozen in their global position in space".

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