[gmx-users] Semiisotropic pressure coupling

[Justin Lemkul]

On 2/13/15 4:00 PM, Piggot T. wrote:
> Hi,
>
> In addition to my previous message, I thought i would also add in my 2
> pennies (cents) worth to the discussion on lipid force fields and cut-off's:
>
> From the fairly large range of force fields and simulation parameters I have
> looked at with simple PC membranes, Chris you are most definitely correct
> that changing the van der Waals cut-off can have a large impact upon the
> membrane properties (e.g. see the results for the Berger parameters in the
> paper I linked to previously and in particular properties like the lipid
> diffusion). Despite this, I would also say that the CHARMM36 force field was
> bar far the most sensitive to small changes in the LJ cut-off. For me this is
> of concern and an area I would imagine there is work currently underway
> (Justin could probably tell us more in this regard). Despite these concerns,
> if you do use the appropriate cut-off's, etc. with this force field I would
> say that it is generally the most accurate in terms of reproducing the
> details of the lipid membrane (e.g. deuterium order parameters, membrane
> thickness, etc.). Given the fact that C36 lipids have faster diffusion than
> many of the united -atom force fields out there and this therefore will
> offset some of the computational costs associated with using this force
> field, I would most likely recommend this if someone asked me which force
> field they should use for a PC lipid membrane. Obviously adding in other
> components, etc. and things get even more complicated but a CHARMM force
> field still probably remains a good choice.
>

Very nice endorsement :)  I should have linked your paper before when asked for 
a URL or reference; I was thinking only in terms of my own work on this topic 
but your efforts are quite nice.  I can't offer anything more detailed than this 
write-up.  I don't directly work on CHARMM lipid force field parametrization, 
though I have collaborated with people who do in terms of getting things right 
in GROMACS, the CHARMM-GUI server, etc.  At the moment, our only real advice is 
"use the settings we're giving you because they're sure to work."  I honestly 
don't know if any other efforts are underway to refine the lipids to make them 
more flexible, but if you've seen any of the guts of how they were parametrized 
and how long it took to get such good agreement with experimental properties, 
you'd probably agree that it would take a pretty massive effort for no real 
payoff when there are working settings in multiple software suites.  We see 
consistent agreement between CHARMM, NAMD, and GROMACS (to name a few) when 
using proper settings.

-Justin

-- 
==================================================

Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

==================================================