Re: Reg : MD simulation with DMSO instead of water

From: Kenno Vanommeslaeghe (
Date: Tue Jun 17 2014 - 13:13:02 CDT

There does exist a thoroughly optimized DMSO model for CHARMM; see
M. L. Strader, S. E. Feller, J. Phys. Chem. A 2002, 106, 1074-1080
It has seen a passable amount of use since its publication, and a very
close variant is present in CGenFF (RESI DMSO).

As for the need to reparameterize the amino acid, strictly spoken, one
could say simulating a transmembrane helix in the very nonpolar
environment of an explicit bilayer also requires reparameterization of the
amino acids, which we're obviously not doing. An unlimited amount of time
can be spent on refining the force field description, and at some point,
one has to concede the potential improvement is not worth the extra
effort, especially if one's goal is to perform an application study. Where
that point lies depends on a number of factors. DMSO is not as far from
water in polarity and hydrogen bonding properties as a lot of other
organic solvents, so the error introduced by immersing amino acids
optimized for water in it would be expected to be on the subtle side.

As for the "water models" argument, biomolecules often display very subtle
conformational behavior "by design" in water, so switching water model is
tempting fate, and the effort of using the right water model is usually
negligible anyway. The effects Akshay is after are likely less subtle than
that (another factor that determines whether to refine the model).

What I'm more worried about is simulating a single amino acid, regardless
of the solvent. Assuming no capping groups are used, the protein FF is not
optimized for free amino acids, and one needs to worry about whether it
assumes a zwitterionic form in DMSO (and whether the side chain is
protonated, where applicable). If it's a non-protonatable residue with
acetyl and N-methyl capping groups, there's relatively little to worry
about, though as always, a critical mind should be used at all times.

On 06/17/2014 07:12 AM, Axel Kohlmeyer wrote:
> On Tue, Jun 17, 2014 at 1:04 PM, Akshay Bhatnagar
> <> wrote:
>> Can anyone please tell how can i solvate an amino acid in DMSO instead of
>> water. Like from where i can get the parameter, topology, pdb and psf files
>> for DMSO?
> like with any other (custom) molecule: from the published literature
> and with a text and/or molecule editor program
> furthermore, you will likely also need to reparameterize the amino
> acid, since the typical parameters are tuned for a specific(!) water
> model. if people already criticize the change of water potentials as
> inconsistent, you can easily imagine what the attitude towards
> changing to a completely different solvent is going to be.
> axel.
>> Thank you
>> With Regards
>> Akshay Bhatnagar
>> PhD Student
>> BITS Pilani Hyderabad Campus

This archive was generated by hypermail 2.1.6 : Thu Dec 31 2015 - 23:20:52 CST