You are here

Score and I_sc in ROSIE symmetric_docking output

12 posts / 0 new
Last post
Score and I_sc in ROSIE symmetric_docking output
#1

If one is doing a symmetric_docking run with 6 sub_units in cyclical symmetry,
is the I_sc value the total for all 6 interfaces or just 1 of the 6 interfaces?
Is the score the total for all 6 sub_units or just 1 of the 6 sub_units?
Does the score already include the appropriate contribution from I_sc?

Also, in each docking2 run, 2 plots are made: "Interface score I_sc / RMSD" and "Score/RMSD".
Why not make both of these plots for each symmetric_docking run as well?

Finally, if one wants to improve a result from a symmetric_docking run,
what would you recommend?

Category: 
Post Situation: 
Sun, 2015-08-09 00:26
AdoCbl

Predicting the native oligomerization state is i non-trivial problem in itself, since one has to consider the free energy of binding and changes in translational and rotational entropy. We discussed this problem for coiled coil in this article http://onlinelibrary.wiley.com/doi/10.1002/prot.24729/abstract;jsessioni.... We found that a simple heuristic is quite reasonable: Simply compare the per subunit Rosetta energies. I have no benchmark result to support that the interface score should be better or worse than the total score, but a similar heuristic should be applicable (divide the value by the number of subunits). The I_sc should be the total for complex. There is no value reported for the inputed monomer, but you should easily be able to calculate the value for the monomer of the final complex model (with optimized side chains for the complex):

I_sc = full_energy - 6*monomer_energy (hexamer case) -> monomer_energy = (full_energy - I_sc)/6.

Sun, 2015-08-16 03:44
IAndre

But work from the paper seems to use 

 full_energy  / n

rather than 

 (full_energy - I_sc) / n

as the criterion. And also asymmetric relax was performed to improve prediciton results (e.g. Figure 3 the prediction using REU/n would be wrong if no asymmetric relax was not performed). 

Thu, 2015-10-29 04:38
attesor

Thanks for the helpful response, IAndre.

Tue, 2015-08-18 18:39
AdoCbl

I recently read the following article:

"Modeling Symmetric Macromolecular Structures in Rosetta3"
by Frank DiMaio , Andrew Leaver-Fay, Phil Bradley, David Baker, Ingemar André
http://dx.doi.org/10.1371/journal.pone.0020450

Its pp.3,5,8,9 have expressions for the energy E for different cyclical Cm complexes,
each with m subunits. Following this article's logic and notation,
I got the expressions below:

C1: E=1*VRT0
C2: E=2*VRT0+1*(VRT0:VRT1)
C3: E=3*VRT0+3*(VRT0:VRT1)
C4: E=4*VRT0+4*(VRT0:VRT1)+2*(VRT0:VRT2)
C5: E=5*VRT0+5*(VRT0:VRT1)+5*(VRT0:VRT2)
C6: E=6*VRT0+6*(VRT0:VRT1)+6*(VRT0:VRT2)+3*(VRT0:VRT3)
C7: E=7*VRT0+7*(VRT0:VRT1)+7*(VRT0:VRT2)+7*(VRT0:VRT3)
C8: E=8*VRT0+8*(VRT0:VRT1)+8*(VRT0:VRT2)+8*(VRT0:VRT3)+4*(VRT0:VRT4)
C9: E=9*VRT0+9*(VRT0:VRT1)+9*(VRT0:VRT2)+9*(VRT0:VRT3)+9*(VRT0:VRT4)

In the above, E is the "score" listed by a ROSIE symmetric_docking job in
its spreadsheet chart and plotted in its "Score/RMSD" plot. Meanwhile, "I_sc"
in a ROSIE symmetric_docking job's spreadsheet chart is the total of all the
(VRT0:VRTn) terms listed above.

The VRT0 terms are internal scores for each subunit, so they do not contribute to I_sc.
The rest of the terms, (VRT0:VRTn) below, contribute to I_sc:
(VRT0:VRT1) terms are for the interactions between adjacent subunits.
(VRT0:VRT2) terms are for the interactions between subunits separated by 1 subunit.
(VRT0:VRT3) terms are for the interactions between subunits separated by 2 subunits.
(VRT0:VRT4) terms are for the interactions between subunits separated by 3 subunits.

For example, if there were 6 subunits numbered from 1-6 clockwise:
(VRT0:VRT1) would be for interactions between subunits 1-2, 2-3, 3-4, 4-5, 5-6, and 6-1.
(VRT0:VRT2) would be for interactions between subunits 1-3, 2-4, 3-5, 4-6, 5-1, and 6-2.
(VRT0:VRT3) would be for interactions between subunits 1-4, 2-5, and 3-6.

Fig.4 on p.5 lets interactions between things more than 10 Angstroms apart be
neglected. This means only VRT0 and (VRT0:VRT1) terms are kept for large complexes.

Does the above all seem right? If not, please suggest corrections to it.

Thanks again!

Tue, 2015-08-18 19:01
AdoCbl

If anyone knows energy expressions like in https://www.rosettacommons.org/comment/8631#comment-8631 but instead for dihedral symmetry complexes, please post them or a link here.

Thanks!

Wed, 2015-09-23 15:18
AdoCbl

I suppose you can run make_symmdef_file_denovo.py  provided by Rosetta for the Dn symmetries and look at the output SDFs.

Tue, 2015-10-27 05:18
attesor

Thanks for your responses, attesor.

Sat, 2015-10-31 14:12
AdoCbl

https://www.rosettacommons.org/docs/latest/application_documentation/doc...
gives the following useful information:

1. total: The total score is an overall measure of the energy of
the complex.

2. I_sc (interface score): I_sc is the total score of the complex
minus the total score of each partner in isolation. Typical
values for I_sc of good decoys are in the range of -5 to -10.

Tue, 2015-08-18 19:04
AdoCbl

Since I_sc and RMSD seem like important values in symmetric_docking jobs,
why not list them closer to the left end of the spreadsheet,
like each docking2 job does?

Wed, 2015-08-26 19:52
AdoCbl

I am a new user of the rosettacommons forum. Will you please tell me how to send to all the users a new post? I cannot find the corresponding button.

 

Brett

Sat, 2015-11-14 21:35
brettliuliu

Go to the main page. (https://www.rosettacommons.org/forum) Click on the name of the appropriate forum for the topic of your new post (the links down the left hand side).  At the top left of the page you arrive at, there should be a "New topic" button, which should allow you to post a new thread.

Tue, 2015-11-24 16:33
rmoretti