[gmx-users] NOE restraints and undefined hydrogen atoms, virtual interaction sites

[Louic Vermeer]

Hi everybody,

I want to do a refinement of an NMR structure, using NOE restraints. I 
want to do this in SDS, using an SDS topology file[1] that was designed 
for gromacs, but there are no explicit hydrogen atoms. I think it's not 
a good idea to simulate a peptide with explicit hydrogens in combination 
with an all-atom SDS topology (correct me if I'm wrong). If I understood 
correctly from a previous (different) question[2], virtual interaction 
sites can be used instead of hydrogen atoms to define NOE restraints.

Now all I need to do is calculate the positions of these sites. I 
thought it might be a good idea to use the "protonate" command on the 
input structure of my peptide (.gro) and calculate the position of a new 
virtual interaction site for the NOE constraint. This new virtual atom 
would have be positioned relative to the atoms that are available in the 
gromacs all-atom forcefield because it will be added to the 
non-protonated input file.

I tried to use "pdb2gmx -vsite hydrogens" command, but it either:
- does not seem to recognise the atom types of the protonated input 
structure because I select the ffG43a1 forcefield for the output file.
- creates only virtual interaction sites on my LYSH residues in a 
non-protonated input file because they are the only ones in my peptide 
that still contain protons in ffG43a1.
This behaviour seems normal to me, but let me know if I'm missing 
something here. Maybe I'm not using pdb2gmx the right way? (more 
information below[3] )


My questions are:
- Can pdb2gmx generate the virtual interaction sites for me? How? And if 
not:
    - Is the procedure described above correct, or is there a better way?
    - Does a script/program to calculate virtual interaction sites based
      on amino-acid protons already exist? I'm capable of writing one
      myself, but since it takes time I only want to do that if nothing
      is available already.
    - Am I correctly assuming that I would add a non-charged virtual
      interaction site without a mass (so only a position to define the
      restraint because the proton mass and charge are already accounted
      for in the all-atom forcefield)


Thanks beforehand, I'm looking forward to my "mdrun"...

Louic Vermeer
IPBS / CNRS, Toulouse
Wageningen University, Wageningen


-- footnotes below --

[1] http://www.apmaths.uwo.ca/~mkarttu/downloads.shtml
[2] http://www.gromacs.org/component/option,com_wrapper/Itemid,165/

[3]

- starting from a pdb-file containing "pseudo atoms"
   # pdb2gmx -f input.pdb -o output.pdb -inter -ignh - vsite hydrogens
   Fatal error:
   Atom QB in residue LYSH 1 not found in rtp entry with 13 atoms
      while sorting atoms

- starting from a protonated file without the pseudo atoms:
   # pdb2gmx -f input.pdb -o output.pdb -inter - vsite hydrogens -ignh
   No error, but only LYSH gets virtual interaction sites. Removing the
   "-ignh" leads to an error again: Atom "HA" not found in residue LYSH.


-- end --