[gmx-users] Problems in the modelling of interaction between peptide and copper

Alex nedomacho at gmail.com
Thu Apr 23 09:18:01 CEST 2015


Your model is not reasonable. First of all, let's forget about
simulations for a moment, because GMX is a wildly inappropriate choice.
Even if the energy was not zero (which in your case is probably due to
some LJ parameters set up improperly, or some really good GMX code
giving you a hint that this is a bad idea), the results would describe a
protein next to a collection of LJ entities put together in a crystal
lattice. Here is why:

When exposed to air or water, copper surface
would undergo quick oxidation. That aside, let us say we have a magic
copper surface, which is not oxidated. This is even more problematic
for MD-type methods, because metal surfaces don't really have
conventional van der Waals due to presence of Fermi gas in
metallic crystals, as opposed to the type of polarization in covalent
dielectrics or even semiconductors.

The only method somewhat reasonable for metals in MD I am aware of is
the Modified Embedded Atom Method (MEAM), which, I believe, is implemented
in LAMMPS. At the same time, it won't handle the protein.

Alternatively, you can use DFTMD in CP2K and actually place
the short region of your protein close to the surface of copper and
see what happens. This in fact appears to be a half-decent idea and
I'd try it. Depending on the exchange correlation functions and the
basis you choose, you could probably do about a 100 atoms for a few
picoseconds within reasonable time with parallelization over 40-50
decent cores. One thing to keeo in mind is that
the last time a checked (a few months back), CP2K had no electron
energy sampling aside from the Gamma point (VASP can do that, but it
is expensive). If that doesn't bother you, I would recommend going the DFT route. You could even try to
build the energy surface for your protein-metal interaction and then
try to use it to parameterize a lower level method such as MD.

Hope this helps.


MT> Dear all,

MT> I got a problem when modelling the interaction of a peptide and
MT> copper. I am using the force field of opls-aa. The charge of
MT> copper atoms are defined as zero and the non-bonded parameters are
MT> obtained from the CU ions, which can be
MT> found in opls-aa.

MT> I applied smd and found that the peptide is approaching the
MT> copper surface and stopped when it is very close to the surface,
MT> due to some abnormal interaction. The whole process seems quite
MT> reasonable, however, when I rerun the
MT> simulation to obtain the interaction energy, for the vdw
MT> interaction between peptide and copper is zero, absolute zero, in the whole process.

MT> Can anyone give me some suggestions? Is my model reasonable? I am
MT> not so confident with modelling strategy of using the LJ
MT> parameters for copper bulk material. Also the zero interaction energy is another big issue.

MT> Cheers,
MT> Tng

More information about the gromacs.org_gmx-users mailing list