[gmx-users] pi stacking

[Justin Lemkul]

On 4/15/15 10:28 AM, soumadwip ghosh wrote:
> Dear Tushar,
>                      I dint get what you mean by 'effect of salt
> concentration can not be simulated in GROMACS'. I am simulating a single
> stranded DNA which is negatively charged in its native state and the added
> salt is bound to affect its dynamics. This is of course valid if the force
> field parameters chosen for the ionic species are correct and the dynamics
> is studied for sufficiently long time in the realm of what we want to
> study. There are numerous examples of protein as well as DNA dynamics
> studied at different salt concentrations using MD.
>
> Dear Justin,
>                     So I guess when I am seeing two base pairs getting
> stacked for a significant duration of time the interaction can be termed as
> 'stacking' whose interaction energies can be calculated by summing up the
> short range non-bonding electrostatic and VDW energy terms. It appears to
> me that this stacking is not a full pi-pi stacking since classical force
> fields can not treat pi electrons or their polarizability explicitly.
>
> Can you suggest me in short what am I supposed to reply to the reviewer's
> query "'what the pi-stacking interactions actually correspond to (in the
> real world) and how well these are captured in the empirical force field
> description?" Should I cite some of the papers (like that of Chen & Garcia)
> to show that Charmm deals reasonably well with non-bonded stacking
> interactions? Can you suggest some more appropriate references? Or any
> other comment regarding the reply of the above review. Thanks for your help
> as always in advance.
>

A few more possible references are mentioned in 
http://www.charmm.org/ubbthreads/ubbthreads.php?ubb=showflat&Number=13436

The empirical force field generally captures the effects pretty well, though 
admittedly not perfectly (no force field is perfect :)  If one looks at ideal 
base stack geometry QM interaction energies 
(MP2/cc-aug-pVDZ//DF-SCS-MP2/aug-cc-pVTZ, a la dx.doi.org/10.1021/jp308364d) 
with those of CHARMM36, the agreement is pretty good.  Unfortunately, none of my 
results are published and are part of ongoing development, so that doesn't 
really help you much.  But across 32 ideal conformers (all 16 base stack 
combinations, both A- and B-DNA geometries), the average difference between 
CHARMM36 and QM is 0.13 kcal/mol, RMSD of 1.67 kcal/mol).  Accuracy depends on 
the nature of the individual base stack sequence.

Of course, you could do some high-level QM on some representative conformers and 
do the MM interaction energy, as well, because with a single-stranded nucleic 
acid sequence you'll have lots of non-canonical geometries, something I'm only 
scratching the surface of now from the development standpoint.  Doing that would 
be pretty good proof.

-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

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