[gmx-users] pdb2gmx not recognizing disulfides

Mark Abraham mark.abraham at anu.edu.au
Wed Jul 4 07:40:15 CEST 2007

> Hello Mark,
> a) I know for sure that the structure has a given number of disulfide
> bonds.
> I have several other Gromacs simulations where the bonds are present
> initially and stay bonded; my current problem has to do with converting
> this
> Amber trajectory to Gromacs format (which I'm doing becuase I'm most
> familiar with gromacs tools right now).
> b)This structure came from a pdb crystal structure orginally. It was
> properly minimized, equilibrated to 300K and simulated in a box of water
> for
> about 10ns.

Fine, but if I remember the discussion accurately, both of the above
didn't have the sulfur atoms that you think should be bonded at a distance
consistent with a single bond. If not, why not?

> c) The structure was already minimized.

That's not the point. If you have sulfurs that are too far for pdb2gmx to
create a disulfide bond, then one way to get them close enough is to force
the minimizer to put them close.

> d) The only usuable structure file from Amber for gromacs conversion is in
> the pdb format, which is what I used. I did use a pdb frame from later in
> the trajectory but I get the same results.

So either you're doing the same thing wrong, or the same disulfide(s)
distances are not consistent with disulfide bonds.

> The structure consists of 4 chains with disulfide bonds in several places
> and some between chains. When I use the -merge option, I get problems with
> the chains. When I use the -ignh option, all the cysteines are protonated
> (which is not what should be happening). If I use the -ss option...nothing
> happens.

It sounds like you should be using -merge and -ss and specbond.dat
correctly on a structure file that has the S-S distances corresponding
approximately to the entries in specbond.dat If not, you will have


More information about the gromacs.org_gmx-users mailing list