You are here

scoring of beta sandwich protein

7 posts / 0 new
Last post
scoring of beta sandwich protein


I wonder if the scoring scheme used in Hu et al. 2008 Structure is availabe in Rosetta 3.5.

I found talaris2013 beta, but it is identical to talaris2013 cart, including comments. Could it be a bug?

Post Situation: 
Mon, 2014-11-17 08:24

in addition, why does -use_bw mean in the command line used in the paper? I could not find helpful anything searching this forum / this website / internet :(
( it seems there is another option called -use_aw )

Tue, 2014-11-18 03:34

Ok, here goes. I can't find a beta sheet patch or scorefunction in the database. The options use_aw and use_bw are commented out - ie they are not used anywhere in the code.

The supplemental does not have the reference energies. I believe these reference energies may have been left in Rosetta++. So, I would first attempt to email the first and second authors for more information. Once you have the reference energies, you should be able to make a new scorefunction with the new references by editing and saving a new weights file like you observe in Rosetta/database/scorefunction/weights. If you get that far, post here for more help - I'm watching this post.

There is a caveat with this, however. First, the references were optimized for either score12 or something else, so you will need to use the score12 weights. In addition, you will need to pass the -restore_pre_talaris_defaults option. Lastly, I'm not sure what changed in score12 between this paper and the score12 we know from Rosetta3. Presumably nothing did, but there is a possibility that it was reweighted. I've never used Rosetta++, so I'm not sure if the default energy function is even called score12. Perhaps Rocco or someone else on here knows. If not, there may be a possibility that the weights just won't work. Perhaps your best bet might be to use Rosetta++ for fixedbb design with these reference energies...

Tue, 2014-11-18 08:41

Actually, the energy was modified quite a great deal more than that. I would totally try Rosetta++, it should work.

Tue, 2014-11-18 08:49

thanks Jadolfbr!
And how about talaris2013_beta.wts ? That name looks suspicious to me :) it seems someone has tried to update scoring weights for beta it in 3.5.

Tue, 2014-11-18 23:51

I don't know, I would think it meant beta version of talaris2013, but I'm not sure.

Wed, 2014-11-19 07:31

It's part of some development work that's going on. The comment when it was added was "added talaris2013_beta.wts - just a copy of talaris2013_cart for now, add arovirt later if all goes well"

I don't believe that the "beta" here has anything to do with beta sandwich proteins or the Hu et al. paper.

Thu, 2014-11-20 10:46