We need additional score parts..

Case number:845813-989376
Topic:Game: Tools
Opened by:Seagat2011
Status:Open
Type:Suggestion
Opened on:Saturday, March 12, 2011 - 03:25
Last modified:Sunday, May 29, 2011 - 22:55

dihedral angles components for backbone: omega, phi, and psi angles, respectively, of backbone

dihedral angles components for backbone: omega, phi, and psi angles components, respectively, of the peptide backbone




Score parts should be, theoretically, additive.

for example:

rebuild #1 - score part (compiled from 12 residues)

clashing : -647.994
packing : 539.824
hiding : -240.275
bonding : 43.485
backbone : 14.940
sidechain : -126.715]

..a quick non scientific analysis would say wiggle sidechains (ws), to resolve both Sidechains and Clashing, and shake (s) would recover the pose, because wiggle (w) nor wiggle backbone (wb) are needed because the backbone score is already positive. However this theory does not pan out in real life.

The correct order of operations, after much trial and error, are: wb + s

I could not understand, for the life of me, why score parts are not additive.

The error was in assuming the score parts were mutually exclusive.

In fact, the Sidechain score component contains an element of Clashing in it.

Solution: Include a score part: backbone and bonding reference, calculated from recent best.

Example 1

Bonding reference ( non cumulative )
cis - 0%
trans - 0%
gauche(-) - 0%
gauche(+) - 0%

Backbone Reference ( non cumulative )
phi - 0%
psi - 0%
omega - 30%

..this indicates only a wiggle backbone (wb) operation would be needed to recover the highest score, because only the omega angle of the backbone has changed.

Example 2

Bonding reference ( non cumulative )
cis - 0%
trans - 30%
gauche(-) - 0%
gauche(+) - 0%

Backbone Reference ( non cumulative )
phi - 0%
psi - 0%
omega - 30%

..this indicates a wiggle (w) operation, or combination wiggle sidechains (ws) and wiggle backbone (wb) operation, would be needed to recover the highest score, because only the omega angle of the backbone, and the trans angle of the sidechain has changed.

Example 2

Bonding reference ( non cumulative )
cis - 0%
trans - 100%
gauche(-) - 0%
gauche(+) - 0%

Backbone Reference ( non cumulative )
phi - 0%
psi - 0%
omega - 30%

..this indicates a wiggle sidechains (ws) operation, followed by a wiggle (w) operation, would be needed to recover the highest score, because only the omega angle of the backbone, and the trans angle of the sidechain has changed - though the trans angle to a greater degree.

The score parts are not additive, but are based solely on the overall degree of change.

Example

If A and B sidechains change, regardless of their magnitudes, the degree of change for both, cumulatively, should be 100% ( because 2 out of 2 changed ) .

(Sat, 03/12/2011 - 03:25  |  4 comments)


Joined: 06/17/2010

I think the "messy" one is clashing part. Because it indicates both sidechain and backbone clashes. If we can broke that 2 in pieces. Also bonding part may be divided to bb and ss bonding. All those changes can improve our scripts :)

Joined: 08/24/2010

Hi, rav :)

Yes, for you me and others who actually utilize these in our recipes, and try to interpret them, I think we would agree the score parts need a revamping. If the values mean nothing they mind as well be pulling these values out the proteins butt.

itskimo's picture
User offline. Last seen 7 years 4 weeks ago. Offline
Joined: 08/10/2010
Groups: foldeRNA

Thank you seagat2011 for such an informitive understanding of folding. I hope this well be saved in the wiki :)

Joined: 06/17/2010

Bump. Is there any chance to make that happen? Or scoring algorithm is so deep that it cannot be done?

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