From: Anton Arkhipov (
Date: Sun Jan 17 2010 - 23:37:48 CST

Hi Francesco,

Sorry for a delay with answering your message. I hope you already
figured this out, but if not, here we go...

My suggestion is that you make SBCG models of your three chains
separately, and then combine them in one PDB/PSF. It's just the same
way as working with an all-atom heterotrimer of proteins: you have
topology file, PDB files for each of your chain, and then you run
psfgen on those to get the PSF/PDB of the trimer. The difference with
SBCG is that you'll need to process your three chains separately first
with the SBCG Builder, and get three SBCG topology, parameter, and PDB
files - one for each chain. Then you put them together using psfgen,
like in the all-atom case.

If you process all your three chains together by SBCG Builder, it will
think that the three chains are all together, like one protein. Then
indeed you may have a situation where the Builder connects the chains
by some bonds.


On Dec 30, 2009, at 1:22 PM, Francesco Pietra wrote:

> Hello Anton:
> Just a quick question. I tried your sbcg with a trimer (three
> subunits, or three chains if you like) protein, created the cg pdb,
> .top and .par files using "Determine bonds from all atom".
> Then, in building the cg .psf file with PSF Builder, the "segments
> identified" were just a line
> 01 1 1- 42 none none Other
> Does that mean that "bonding" was also created between the three
> chains, arriving at a sible chain? From the graphic screens it looks
> like so.
> If my guess is correct, what about building the sbcg model separately
> for all three subunits and then combine the results? Is that possible?
> Thanks and happy new year
> francesco pietra