Re: [AMBER] ion parameters in simulations with nucleic acids

From: Vlad Cojocaru <>
Date: Tue, 10 Apr 2012 23:51:57 +0200

Dear Tom,

Thank you very much for your response ... Yes, of course you are right that the paper does conclude mainly that simulations are in general independent of the ion model and that all models do a reasonable job at low concentrations of ions.

If we will find some spare time and resources maybe we'll try to simulate with different ion parameters to see if there is an effect.

In the search for the best protocol for our simulations, I wanted to see if anybody has some other test simulations with DNA and different ions that might argue for one or the other model ...

Thanks for Dang's parms... Of course its easy to extract them from literature and in fact we did that . ... I was probably a bit lazy when reading all the ions papers and posting my message I hoped that somebody had the files already .....

Best wishes
Dr. Vlad Cojocaru
Max Planck Institute Muenster
Roentgenstrasse 20,
48149 Muenster, Germany
Tel: +49-251-70365324
Sent from my mobile phone; Sorry for being short and for any errors
Thomas Cheatham III <> wrote:
> Since the new parameters from Tom Cheatham's work were added to AMBER we 
> used those parameters in the simulations. However, recently I came 
> across a paper coming from Modesto Orozco's group (PCCP 2009), showing 
> that these parameters are less suitable than for examples Dang's 
> parameters in simulations with nucleic acids.
I am not sure that paper quite says that as they point out: 
"...we also show that DNA simulations (at least performed with AMBER 
force fields) in near-physiological conditions are quite robust and not 
much dependent on the specific details of the ion model."
"In summary, the structure of DNA duplex (at 500 mM salt concentration 
is independent on the ion model and even the most risky ion models could 
be safely used to obtain structural descriptions of DNA duplexes."
Other claims in the paper, such as over-neutralization of phosphates and 
that NaCl is preferred over KCl, are not strongly justified as I cannot 
see any statistically significant difference in the DNA structural 
My opinion is that AMBER DNA is showing too little salt dependence and 
that likely, you will not see significant differences in DNA duplex 
structure/dynamics with different monovalent ion models (unless of course 
a crystal crashes into the DNA which can destroy the structure). The same 
may *not* be true with quadruplexes where ions can associate in loops; 
none of the ion models we have applied accurately retain ions bound into 
the loops, so with non-canonical structure your mileage may vary.
If you do see (statistically significant) differences with various 
monovalent ions on DNA structure I'd be interested to hear about it.
> Also, I would like to ask if somebody has the leap libraries and 
> parameters for Dang's parameters. I could not find them neither in the 
> AMBER distribution nor in the contributed parameters. I hope I am not 
> too blind here ...
It is not too hard to pull them from the literature, however this is what 
I've used in the past:
smith/dang ion parameters
IM 35.45 3.235 assumed to be Cl- (ion minus)
IP 22.99 0.250 assumed to be Na+ (ion plus)
Na 22.99 0.250 Na+, ions pol:J.PhysC,11,1541,(1978)
Cl 35.45 3.235 assumed to be Cl- (ion minus)
K 39.10 1.060
IM 2.47 0.1 Smith/Dang
IP 1.32 0.13 Smith/Dang
Na 1.32 0.13 Smith/Dang
Cl 2.47 0.1 Smith/Dang
K 1.8687 0.1
AMBER mailing list
AMBER mailing list
Received on Tue Apr 10 2012 - 15:00:10 PDT
Custom Search