AMBER Archive (2005)

Subject: Re: AMBER: Targeted MD

From: Viktor Hornak (hornak_at_csb.sunysb.edu)
Date: Wed Nov 30 2005 - 11:46:55 CST


Dear Whitney,

Whitney Allen wrote:

> Dear Amber users,
> I would like to know the limits of targeted molecular dynamics. I have
> generated a pathway for an RNA molecule that consists of several
> intermediate structures. When I go from one structure to another that
> has an all-atom RMSD of 2.5 angstroms it converges to the reference
> structure with little perturbations. When I change calculation so that
> the structure goes to another reference structure that has a
> higher all-atom RMSD, say 8 angstroms, it does converge to the
> appropriate structure, but when I plot the RMS from the first
> structure in Ptraj I get an oscillatory damping behavior (similar to
> an underdamped mechanical system). I am using pretty small force
> constants tgtmdfrc=0.01 to tgtmdfrc=0.1 (I have tried higher values
> but it doesn't change the results) and have the tgtfitmask and
> tgtrmsmask set to include only the backbone. So now to! my questions.
> First, if 8 Angstroms is too high of an RMSD, what is considered a
> reasonable maximum value? Second, what causes the oscillations, I have
> tried multiple variations of tgtmdfrc, tgtmdrms, and RMSD values
> between intermediate structures and still get this pattern. Little is
> written on how the forces generated from targeted MD are weighted in
> with the other forces used to create a trajectory. Since the molecule
> does converge to the correct structure, is this oscillatory behavior
> considered to be okay? Please help me understand the basics of
> targeted MD. Below is a sample input code.
> Thanks
> Whitney
>
> cat <<eof > mdin
> equilibrate the structure
> &cntrl
> imin=0, ntx=5,
> irest=1, ntxo=1,cut=15.0, tempi=300.0,
> ntpr=100, ntwx=100, ntwe=1000,
> nstlim=2000000,tautp=2, temp0=300.0,
> dt=0.001, nscm=100,
> igb=1,ntb=0, saltcon=0.5, gbsa=1,
> ntt=1, nsnb=20,offset=0.13,
> itgtmd=1, tgtrmsd=0.5, tgtmdfrc=0.01,
> tgtfitmask=":1-76_at_P,O1P,O2P,O5',O4',O3',O2',C5',C4',C3',C2',C1'",
> tgtrmsmask=":1-76_at_P,O1P,O2P,O5',O4',O3',O2',C5',C4',C3',C2',C1'",
> &end
> Group for target
> RES 1 76
> END
> END
> eof

you're basically trying to change your initial structure to the target
one in "one big RMSD step" even though you allow 2ns for it. For larger
RMSDs between initial and target structures the forces may get too big
and may cause distortions of your molecule and possibly the oscillatory
behaviour you describe due to periodically refitting and forcing the
selected portion of the molecule. For simulations of gradual changes
from one conformation to another I suggest using weight changes on
TGTRMSD and gradually decreasing initial RMSD value (of your initial
structure to reference, i.e. target structure) to 0.5 (or something
close to zero). Check the example input in
$AMBERHOME/test/tgtmd/change_target.rms
If part of your molecule is not changing, you can even use positional
restraints (ntr=1) in conjunction with your TMD restraints (tgtrmsmask).
Check the example input in $AMBERHOME/test/tgtmd/change_target.ntr.
I am not sure what the intended purpose of the group input at the end of
your input is ("RES 1 76", etc). As far as I know, it shouldn't even be
read in.

Cheers,
Viktor

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