814 RMSD filter improvement

Case number:699969-996384
Opened by:bob1928
Opened on:Monday, December 2, 2013 - 03:34
Last modified:Tuesday, December 10, 2013 - 23:01

From 814 description:
"there is an RMSD condition on most of the residues, to maintain the ferredoxin-like fold"

In addition to maintaining the fold, the filter is also enforcing the macro protein position and rotation. I don't think that was the intent?

Some scripts (GAB BiS especially) tend to rotate the entire 5 protein group about the symmetry axis. This can cause condition violation even though the individual protein fold is almost unchanged. In all cases encountered so far, the condition can be re-satisfied with manual moves (strictly translate and rotate not altering fold). Manual realignment to the guide can waste a great deal of time.

I would like to suggest there should be a function very much like "align guide" to move (translate and rotate only) the model used as the RMSD reference.

foldit_1385951478.png289.6 KB
(Mon, 12/02/2013 - 03:34  |  3 comments)

MurloW's picture
User offline. Last seen 5 hours 12 min ago. Offline
Joined: 11/21/2012

I also believe this was not supposed to happen. A way of fixing things after BiS recipes is freezing everything, dragging it to the RMSD site and wiggling it up and down there before unfreezing. Nevertheless this is still a waste of time since we shouldn't need to do that anyway. I'm quite certain we're missing out on some superior sidechain arrangements since most folders will just restore credit best after a long recipe, thus reverting to probably the very beginning of a BiS recipe getting gains. Lots of time and resources wasted.

bkoep's picture
User offline. Last seen 9 hours 32 min ago. Offline
Joined: 11/15/2012
Groups: Foldit Staff

For Puzzle 814, we did indeed want to preserve the orientation of each subunit relative to the symmetric axis (i.e. restrict translation and rotation). Sorry, this could have been more clear in the puzzle description.

We may repost this puzzle in the future with an "aligned" RMSD condition like you have suggested.

katfish's picture
User offline. Last seen 5 years 50 weeks ago. Offline
Joined: 01/09/2013

We unfortunately haven't had any luck reproducing the issue with 814 internally, but we are continuing to investigate nonetheless. You can help us out a lot by continuing to keep an eye out for anything similar going forward. You've done a bang up job so far, and we couldn't have gotten this far without your help!

The positive development to come out of this is that we do have a new crash reporting feature in development that should help us chug through bugs that are tricky to reproduce in-house.


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