AMBER Archive (2009)

Subject: Re: [AMBER] Problem with RNA hairpin unwinding in explicit water

From: Thomas Cheatham III (tec3_at_utah.edu)
Date: Wed Apr 22 2009 - 17:56:25 CDT


> I'm trying to simulate a hairpin structure of RNA using an NMR structure
> as a starting point (1MFY in PDB). However, the RNA unwinds from a
> helical structure into a linear base-paired one. This happens within
> several ns of simulation.

Anything this drastic happening so fast is a little suspect; normally we
can simulate hairpin loop geometries with ff99 or ff99-bsc0 for 5-10's of
ns without significant change. 50-100 ns is another story.

- If you are restarting frequently, it could be the random seed problem
with Langevin (see paper by Cerrutti/Duke and discussions in archives).

- I do not think no salt would cause structure disruption and certainly
not on this time scale as Ewald implicitly includes an effective salt
(i.e. the net-neutralizing plasma). Years ago we looked at this in short
simulations and saw little salt effect. However, perhaps as the defaults
now get rid of excess charge, the problem could be related to
net-neutralization of the overall large negative charge (thereby changing
each of the charges). You would see this at the top of the mdout file,
i.e. removing net-charge of X. But, if you have a lots of atoms, the
change in charge is only +31/#atoms which is likely small. You could try
to turn on the Ewald plasma (i.e. don't neutralize) but better would be to
include a finite salt concentration. Somehow I do not think this is the
origin of the problem.

- If you see sugars systematically in the plane of the base (alternating)
this is indicative of the alpha/gamma problem that ff99-bsc0 intends to
overcome.

- Note that with Ewald a cutoff of 15 is serious overkill and will cost
your dearly in simulation time. I would recommend 8-9 angstroms with a 1A
skin and the uniform van der Waals density correction (which is on by
default).

- Check to make sure your structure post-equilibration is OK.

- You can probably get away with DT=0.002 (set the SHAKE tolerance
tighter, i.e. smaller).

--tec3

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber