Re: Dioxygen atom type with CHARMM36

From: Kenno Vanommeslaeghe (
Date: Tue Nov 11 2014 - 12:46:48 CST

The bond length and force constant is trivial; see Alex MacKerell's force
field papers. The problem here is the L-J parametrization, which, as
detailed in the same papers, is usually based on reproducing liquid
densities, heat of vaporization and free energy of solvation. The latter
seems particularly important in the present case because parameters
optimized on cryogenic LOx might not transfer well to room temperature,
especially considering that it contains dimers. Unfortunately, targeting
solvation free energies in parametrization also happens to be particularly
laborious and time-consuming, which, together with the relatively uncommon
occurrence of free oxygen in biomolecular MD simulations, is why it hasn't
been done before in a CHARMM context. If you're unhappy with that, feel
free to do it yourself.

PS. in the hypothetical case someone here decides to go ahead with it,
please don't forget to set the multiplicity to "3" in any and all QM
calculations! It's incredibly easy to overlook.

On 11/11/2014 12:19 PM, Francesco Pietra wrote:
> Thanks.
> Could you suggest a protocol to validate params for aerial (i.e., triplet,
> if that matters) dioxygen in water solution? It is high time (in my view)
> to do that. Spending much computer time on (or raw) uncertain basis is
> something curious in 2014. It seems to be a situation common to all ff for MD.
> thanks again
> francesco
> On Tue, Nov 11, 2014 at 6:07 PM, Kenno Vanommeslaeghe
> < <>> wrote:
> I don't have a clear-cut answer to this question, but can offer some
> food for thought:
> <>
> <>
> Bottom line is, as always, that it all depends on what kind of
> accuracy you need. I certainly wouldn't use it for any kind of
> quantitative (e.g. free energy) calculation without stringent
> validation, but if you just want qualitative sampling information on a
> system in which the oxygen does not play a central role, a very
> approximate model that is either based on "OM" in
> toppar_all36_prot_heme.str or OG2D1 in CGenFF may be good enough.
> On 11/11/2014 11:11 AM, Francesco Pietra wrote:
> Hello:
> My question is about the heroic use of dioxygen data from
> toppar_all22_prot_heme.str (by zeroing the charges) and
> par_all27_prot_lipid.rtf for CHARMM36. I refer to free dioxygen
> in TIP3
> water solution.
> With CHARMM36 the "OM" type of previous FF version should be
> changed. Is
> the change to "OG2D1" the best compromise?
> Thanks for advice
> francesco pietra

This archive was generated by hypermail 2.1.6 : Thu Dec 31 2015 - 23:21:21 CST