[gmx-users] Suggestions on running simulations of very long polypeptide chains

[Nash, Anthony]

Hi all,

I¹m looking to run MD simulations of regions of a collagen molecule. A
whole collegen molecule is made up of three polypeptide chains, each
around 1000 residues long (gross generalisation as there are around 24
different collagen protein families). I am only interested in modelling a
section of 30 residues in length, and then as a dimer (30 per chain, three
chains per molecule, 2 molecules to make the dimer, 180 residues in the
whole system)

I tried looking at the structure as a free-floating dimer in solvent. Two
things occurred, 1) they became monomeric, 2) the system box had to be
over twice the size in all three dimensions to consider all of the
possible conformations without seeing its period image. This went from
being a very small system to a very large system due to the solvent.

After some careful consideration I think my best approach would be to turn
each 30 long residue into a polypeptide chain which covalently binds back
onto itself, giving the illusion of a very long polypeptide chain.  Are
there any tutorials of best practice on this techniques, and in particular
how to manage the pressure on the adjusting volume?

Many thanks
Anthony

Dr Anthony Nash
Department of Chemistry
University College London