[gmx-users] re: Phosphate esters

[Justin A. Lemkul]

Ben Ahmady wrote:
> Dear GROMACS users/developers
> 
> I've recently started a PhD and am currently looking at bilayer 
> formation of two-tailed alkyl phosphate esters. I've been using the 
> PRODRG server to generate topologies for use in GROMACS, and as per the 
> recommendations made in the paper by Lemkul et. al I have consistently 
> altered the charges/charge groups according to the charges reported in 
> similar molecules for which topologies are well established. However, 
> the "similar molecules" to which I'm referring to are phospholipids such 
> as DPPC rather than the alkyl phosphate esters in which I'm interested, 
> and as such I was concerned that the charges [for the alkyl phosphates] 
> I was implementing might not be entirely accurate. I therefore set about 
> rather ham-fistedly doing Bader analysis on the individual alkyl 
> phosphate molecules I'm studying to try to establish more realistic 
> charges to implement in the topology (more accurately: I ran the same 
> simulations concurrently with the established molecules like DPPC and 
> tried to find a relationship between their Bader charges and topologies 
> which I could use to make a realistic estimate of the charges on the 
> alkyl phosphate molecules).
> 
> I really have two questions and hoped that I might take the opportunity 
> to ask experienced users/scientists: the first is whether I'm wrong and 
> that in your opinion(s) it is in fact sensible to simply use the charges 
> from, say, DPPC in my alkyl phosphate molecules, and the second question 
> is - regardless of the answer to the first question - whether what I'm 
> doing is in your opinion(s) sensible. I hope these aren't inappropriate 
> questions.
> 
> 

Parameterization is a tough task, if done properly.  If you find it came easy, 
it's probably not right ;)  It sounds like you're trying to work backwards from 
PRODRG and correct the charges.  While that can be OK, if you've committed to 
using the Gromos96 force fields, I'd suggest you work the other way around. 
Look into the literature and see if anyone else has done anything similar - what 
force field did they use?  Was it successful?  Frankly, the details of 
parameterization for Amber, CHARMM, and OPLS are significantly more detailed 
than those of the Gromos96 force fields, so you may actually find it easier to 
use the QM methods described for those other force fields.

The next most significant issue is validation.  You have to have some criteria 
by which your new parameters are judged.  Again, this is a reason why Gromos96 
can be tough, unless you have DeltaG of solvation for your compounds, or 
suitable models thereof.  There are alternatives, but they're not significantly 
better.  Reproducing some condensed-phase criteria is common to all the force 
fields (some have additional QM geometry requirements).

-Justin

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

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
MILES-IGERT Trainee
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

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