Exploration filter considered buggy

Case number:671071-2000334
Topic:Game: Other
Opened by:Timo van der Laan
Opened on:Friday, February 6, 2015 - 12:21
Last modified:Sunday, February 8, 2015 - 08:49

The way the exploration filter is implemented is not good.
I had a filter score of 2.5 but with just moving the protein by hand I got it back to 1.0.
It should not be computed based on the starting position in space, but it should find the best match or there should be a way to let the protein find the best position with a new button and/or Lua function.

(Fri, 02/06/2015 - 12:21  |  2 comments)

Susume's picture
User offline. Last seen 4 hours 19 min ago. Offline
Joined: 10/02/2011

I agree that comparing the protein with the starting position is a problem. Many of our scripts cause the protein to wander from its starting coordinates, even if they do not change the overall shape much. GAB BiS is one such script; it uses bands in space to stretch the protein and causes a lot of wandering. A better solution might be to align the starting shape to the current shape (but not the other way around) before calculating RMSD. I suppose this would make the filter slower.

As Timo says, a lua function and corresponding GUI tool to align the protein to the starting position would also work. If the player or script controls when this happens, it is ok to align the protein to the original starting position - then we are in control of when any bands to space we might have are made nonsensical by the protein moving. If the game is going to do the aligning without user control, don't move the protein in space, but align the invisible starting structure to the current location of the protein.

For now, if you have a score that is not getting credited due to wandering away from the starting location, save it, reset the puzzle, save the reset as "starting position" or something similar, undo the reset (or load your save), then load the "starting position" save as guide. Freeze the whole protein, then manually move it to overlap the guide to try to get the exploration multiplier low enough to get credit for the score. Do not use the Align Guide tool, as this moves the ghost to your protein, rather than moving your protein to the ghost, which is what you need.

Joined: 05/19/2009
Groups: Contenders

In facr, if you click on the checkbox next to the 'exploration' condition, it will show you the outline of the original position, so you can simply drag it back into that outline.


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