Re: [AMBER] Force fields for DNA, what to choose?

From: David A Case <>
Date: Thu, 2 Jun 2011 11:40:36 -0400

On Thu, Jun 02, 2011, Anna Reymer wrote:

> I have this question because I have performed a couple of 10ns runs with
> ff99 and ff10 parameters of just DNA duplex of 16 bp long

For DNA, "ff10" is the same as "ff99bsc0", so reports in the literature on the
latter should be helpful. See, e.g. Lavery et al, Nucl. Acid. Res. 38:299,
2010, as well as references 22 and 29 cited there. Also: Perez, et al.
NAR 36:2379, 2008.

Tom Cheatham can probably chime in on ff99 results, although my guess is that
they should be pretty similar to the ff94 results, e.g. as shown in Dixit et
al, Biophys J 89:3721, 2005.

> On the other hand, the ff10 force
> fields set provides smaller rise (bp to bp distance) of 2.8 , while
> the ff99 gives the expected value of 3.36 .

The ff10 results are a surprise, given results of Fig. 1 of the Lavery NAR
paper. Does your sequence have lots of YR steps? Then this difference
might be expected (see, e.g. also Fig. 14 of the BJ paper), but the
magnitude of the difference seems pretty large. Of course, 10 ns might
not really sample what you want.

> neither ff99 nor ff10 provides reasonable sugar puckers values...
> the ff99 set gives much more stable values for the sugar pucker then
> ff10, which just covers all possible conformations.

Odd. The sugar pucker distribution of ff99bsc0 (ff10) reported by Perez
et al. shows a distribution whose average is a bit too low, compared
to NMR and Xray results, but qualitatively looks very much like what
is seen experimentally. There is certainly nothing like "all possible
conformations" in their reported distributions; rather, one sees a fairly
sharply peaked distribution, averaging about 135, with only a minor (North)
population outside the range 100 to 180. Again, Tom may know more about
the puckers than I do.


AMBER mailing list
Received on Thu Jun 02 2011 - 09:00:03 PDT
Custom Search