AMBER Archive (2006)

Subject: Re: AMBER: Which one is the best force field for DNA?

From: Thomas Cheatham (cheatham_at_chpc.utah.edu)
Date: Fri Apr 14 2006 - 12:20:14 CDT

> Now, what about ff03? According to your opinion, is it recommended to be
> used in protein-nucleic acids complex simulations? Or is it "safer" to
> stick to ff99 for that purpose? It would be interesting if anybody has
> done simulations of protein-RNA or protein-DNA complexes with ff03 and
> compared them to simulations done with ff99..

Just remember that newer doesn't always mean better :-) I have plenty of
new applicances at home that work fine for the first year or so, but once
the warranty runs out (i.e. you run simulation long enough) they break
down. Yet, our old toaster still works perfectly fine even though it
lacks integrated circuit boards and fancy LCD display...

A caveat with the ff94->ff99 force fields is that issues creap in after
long use. It is well appreciated now that ff94 overstabilizes
alpha-helices, ff96 overstabilizes beta, ff99 slightly overstabilizes
alpha-helices, ... With DNA, ff94->ff99 seem to have subtle misbalance in
the alpha/gamme backbone torsions with DNA that lead to significant
populations of anomolous substates in long simulation. RNA, as mentioned
by Jiri, seems to perform well in simulations on the 50-200 ns time scale.

I personally do not have experience with ff03 so someone else will have to
chime in, but will note that this was developed with proteins in mind.

If I were to do a protein-RNA simulation with what is generally available
today, I would likely use one of the modified phi/psi protein parameter
sets by the Simmerling group (see the JACS 2002 article on TrpCage
folding); if you have amber9 (or are getting AMBER9), this is the
frcmod.ff99SB. I haven't even thought of trying ff03 for protein with
ff94 for DNA for two reasons (1) the files loaded may not be mutually
exclusive as far as atom types, names, etc such that you can not easily
mix them (but I haven't tried it) and (2) ff03 was designed for proteins.

--tom

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu