about rotamers and atoms

Case number:699969-1999026
Opened by:BitSpawn
Opened on:Tuesday, December 30, 2014 - 12:08
Last modified:Tuesday, December 30, 2014 - 15:40

It will help if the order of rotamers is fix to positions of sidechain. On this way we can check good/bad rotamers, and discard bad positions. Now the last rotamer is always (or almost ?) the last best position. at different times the same rot specifies different positions.
As not always there are the same number of rotamers, I think we can have the max positions, and function SetRotamer returns boolean to tell us if that position is possible.

Also, it's a problem that GetAtomCount gives different numbers depending of it has bonds.

(Tue, 12/30/2014 - 12:08  |  2 comments)

brow42's picture
User offline. Last seen 1 year 36 weeks ago. Offline
Joined: 09/19/2011
Groups: None

I think BitSpawn wants to be able to move a sidechain through a fixed set of distinct physical positions. Sidechains tend to bend in certain ways, and it would be nice if we were testing the same one each time. This is so we can loop through them. BitSpawn has noticed the most recent pose (or one similar to it) appears as the last rotamer. If the order changes, then that is duplication of work when we loop through them.

First, though, I'd like a dev/scientist to explain some things about rotamers:

1) How are they selected? I have made a guess here http://fold.it/portal/node/996019
2) Is the current pose one of the rotamers? I have always found a score change when I loop through the rotamers. But, what is the explanation for the issue in the linked feedback (still open).
3) Are the rotamers sorted by probabilty? Are the last rotamers likely to have the same pose? Again, that would be annoying if we are brute-force optimizing the sidechain poses.
4) And finally, can we pretty please have shake accuracy back? Only this time, tell us what they do? Arganine and Lysine have a *lot* of rotamers to check. If you don't want to put them in behavior, you can re-enable the Lua functions and we'll provide the interface and explanations.

Maybe we can have the random sampled rotamers, and the fixed rotamers, though different interfaces.

Joined: 08/09/2010

Thank you Brow, that is :)


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