A few feature requests

Case number:845813-986320
Topic:Game: Tools
Opened by:infjamc
Opened on:Friday, June 19, 2009 - 18:02
Last modified:Thursday, January 5, 2012 - 06:29

Working in the Quest to the Native puzzles gave me a few ideas of tools that would be useful:

1. Loop modeling-specific tools. For example, the ability to cut open a loop at a certain point and then rebuild from both sides would be very helpful when rebuilding long (>8 residues) loops. In QttN5, there is a loop that's ~30 residues long, and matching the backbone residue-by-residue would be difficult due to the constraint that every residue must be kept together. (If anything, this is already implemented in the Protein Design tutorial and Design Puzzle 1 as a "constraint"-- a gap could simply result in a more negative backbone score.

2. A more intuitive way to tweak side chains. As things now stand, one has to hope for the best when tweaking larger sidechains (e.g. Lys, Arg). The ability to, say, freeze part of the side chain and rotate across a certain bond would be useful. (This also raises the question: why not include the option to rotate a sidechain whatever way you want without regard to the rotamers? E.g. have an option to turn rotamers on or off.)

3. One of the techniques actually used in protein structure prediction is fragment insertion, i.e. copying the configurations of short (e.g. 3 or 9 residues) fragments from known structures onto the decoy you're working on. Now, I know that this is already available in all-hands puzzles. But I was thinking if a similar concept could be applied more generally for local rebuilds:

For example, suppose that rest of the protein looks good except for a short, 2-residue loop section. It is too cramped, and you want to expand it. Local wiggling the segment probably won't help much, and neither would bands of strength 10, as the goal here is to either straighten the segment or compress the segment while keeping the rest of the protein fixed. (The closest thing one could do is to unfreeze nearby residues as well, or setting the segment to sheet and then straightening. But that might affect nearby residues as well.)

==> But fragment insertions either using your own solution or your teammate's solution (which switches you to evolver mode) could help here, as would the ability to cut the loop open, drag away part of the protein, and the rebuilding the loop back together. Alternatively, the ability to manually enter torsion angles for a residue might help (yes, I know that doing so would affect everything downstream, but sometimes you do want that to happen).

4. The ability to apply different display options to different parts of the protein. Take QttN5 for example: the starting configuration is quite off. If you want to work only on a certain part of the protein, then having the guide being shown for the entire protein is a major distraction.

5. The ability to save a set of pieces, e.g. backbone and/or sidechains of certain residues, and then calling that set at any time for a function. To take a few examples:

( a ) A residue is in the wrong conformation according to the guide. For some reason, however, it will always flip back to that wrong conformation when a shake is applied. That means manual freezing after every unfreeze is necessary, which is annoying at times. But the ability to apply freezing to a set reduces the number of steps from one for each residue to 2.

( b ) Pressing G aligns the guide to the protein as a whole. But sometimes you might want to align the guide to a certain part of the protein only, which requires right clicking and then choosing align guide. The problem is, that option isn't available while you're wiggling, which is known to affect backbone configurations. As things now stand, the only workaround is to stop the wiggle halfway and then manually apply the align guide. But if the default align guide function (G key) could be manually reconfigured to apply to a set of residues when the set is defined... problem solved.

(Fri, 06/19/2009 - 18:02  |  1 comment)

infjamc's picture
User offline. Last seen 15 hours 38 min ago. Offline
Joined: 02/20/2009
Groups: Contenders
Status: Open » Closed

Closing old case. Interestingly, practically all of these were addressed over time.


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