AMBER Archive (2007)

Subject: Re: AMBER: ff99 vs ff99SB for RNA

From: Thomas Cheatham III (tec3_at_utah.edu)
Date: Thu Jun 14 2007 - 11:17:46 CDT

> I believe this is a pretty simple question about recent force fields. In
> looking through the ff99SB frcmod file, it looks to me like these dihedrals
> should not have any effect on DNA or RNA. But, I recall Carlos mentioning
> something about a small difference in DNA behavior between ff99 and ff99SB on
> this list recently.
>
> Am I remembering Carlos's statement incorrectly, or am I looking at the frcmod
> file incorrectly?

Everyone who has replied is correct; what Carlos was probably referring to
was the small differences in ff94->ff98->ff99 which tweak the chi and
sugar puckering a bit to be more in range. For RNA, those changes appear
to have little influence; with DNA all of those listed (ff94/ff98/ff99)
have problems with the backbone in long simulations so the parmbsc0 is the
current recommended for DNA. For RNA, the jury is still out, but either
parm99 or parmbsc0 are appropriate and compatible with parm99SB and the
RNA should be fine in short runs (~< 20 ns). Note that to use parmbsc0
you need to load not only the modified parameters (after loading the
leaprc.ff99SB) but also the library file that redefines the type of C5';
you can check if your prmtop is "correct" by looking with rdparm to see
that C5' was indeed set with an atom type of CI (with the atoms @C5'
command) as the CI atom type does not exist except in the parmbsc0 force
field.

-- tom

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