Targeted Molecular Dynamics
The Targeted Molecular Dynamics (TMD) method introduces a holonomic
constraint that reduces the rmsd with a predefined target with a
preset value at each MD step (J. Schlitter, M. Engels, P. Kruger,
E. Jacoby and A. Wollmer, Mol. Sim. (1993) 10, 291-308). The method
is implemented for the LEAP integrator, and Berendsen's thermostat
must be used.
* Menu:
* Syntax:: Syntax of the dynamics command
* Description:: Description of the keywords and options
Syntax for the Targeted Molecular Dynamics commands
TMDInitialize { [ INRT integer ] } [ atom-selection ] [ atom-selection ]
{ [ DINC real ] }
{ [ FRMS real ] }
{ [ ZETA ] }
{ [ CZETa real ] }
{ [ ZTOL real ] }
{ [ ZMIT integer ] }
INRT 1000 Number of step that one needs to get rid of artificial
rotational motion in TMD simulation.
DINC 0.0 RMS increment in TMD simulation.
FRMS 1.0D-6 Stop dynamics when a rmsd of FRMS with the target is reached.
ZETA Indicates Zeta form of TMD is being used (default: not active)
CZETa 1.0 Zeta form expotential factor.
ZTOL 1.0E-10 Tolerance used in calculating Zeta-TMD constraint.
Positions are solved for (Zeta-Zeta0) < ZTOL.
ZMIT 1000 Number of iterations allowed in the Zeta-TMD constraint
subroutine.
1st atom-selection Apply TMD fit to the selected atoms only.
2nd atom-selection Apply TMD perturbation to the selected atoms only.
Description of the Targeted Molecular Dynamics Commands
This is only for TMD using holonomic RMS constraint.
For doing Targeted Molecular Dynamics (TMD), one needs to define
a moving coordinate and a target coordinate. You slowly pull the
moving structure towards the target structure by gradually decreasing
the RMS distance between two. The 'pulling' speed is defined
by user.
To invoke TMD, the TMDInitialize command should be given before the
DYNAmics command. After TMDInitialize, the target structure should be
read in with the "READ COOR TARG" command. All TMD variables and the
target coordinates are cleared after the DYNAmics command; before
a restart you should invoke the TMDInitialize command (and the "READ
COOR TARG" command) once again.
To constrain dynamics between two target structures (between a
starting and ending structure of a conformational transition,
for example), the 'Zeta' form of the constraint function is used:
Zeta(t) - Zeta0(istep) = 0 (contraint),
where
Zeta(t) = -1/(1+EXP(-CZETA*RMSD1(t))) + 1/(1+EXP(-CZETA*RMSD2(t)))
Zeta0(istep) = Zeta0(istep-1) - DINC
RMSD1 = (mass weighted) root-mean-squared difference (RMSD)
between the current structure and TARG (using atoms
defined by second selection in TMDI)
RMSD2 = RMSD between the current structure and TAR2
The two target structures are read using READ COOR TARG and READ COOR TAR2,
respectively. The sign of DINC (incrementation of the contraint function)
determines which structure the molecule is pulled towards, and which one is
pushed away from, during dynamics run. For DINC > 0, TMD pulls the molecule
towards TAR2. For DINC < 0, TMD pulls the molecule towards TARG.
The starting value of Zeta0 is based on the coordinates at the start of
the dynamics run; istep is the current step number in the dynamics run.
The ZETA keyword must be used in the DYNA command line for this. If two
targets are read in, but ZETA is not specified, then only the one TARG
structure is used in the TMD algorithm and the Zeta form of the
constraint is not used.
The Zeta form is useful, since it is more effective at pulling molecules
towards target structures than other relative constraint forms, such as
((RMSD1 - RMSD2) - rho) = 0, where the difference in RMSDs may be well
defined, but the current structure may be far from both target structures.
Also, transitions are not limited to paths which only allow for the RMSD
to one target structure to decrease monotomically.
Two atom selections are used with the TMDInitialize command. The first
selection is used to define the atoms used in fitting both targets
to the current structure (done every INRT steps). The second selection
is used to define the atoms which the TMD constraint is applied.
If only one selection is given in TMDI, this selection will be used for
both fitting and applying the constraint. To run TMD in its original
form, one must use 'select all end' as the second selection.
ZTOLerance and ZMITerations are used in the minimization scheme for
calculating the coodinates which satisfy the TMD constraint. They are
similar to the cooresponding terms in the SHAKE algorithm.
The constrained RMSD for one-target TMD is not allowed to go below zero.
Similarily, the restriction |Zeta| <= -1/(1+EXP(-CZETA*RMSD0))+1/2 is
is used, where RMSD0 is the RMS Difference between the two target
structures. Once these values are reached during dynamics, the contraint
value for RMSD (or Zeta) is held at this limiting value.
This subroutine outputs RMSD1, RMSD2, and the actual Zeta value (which
is within +/- ZTOL of Zeta0(istep)), with PRNLEV >= 5. For each dynamics
step, this is likely to print out a few times, due to the iterative scheme
used between this subroutine and the SHAKE subroutine.
Please note that the "standard" TMD runs is parallellized; the 'Zeta-TMD'
has not been parallellized.
Examples: tmdtest32.inp (serial & parallel), and tmd_zeta.inp (serial).
NIH/DCRT/Laboratory for Structural Biology
FDA/CBER/OVRR Biophysics Laboratory
Modified, updated and generalized by C.L. Brooks, III
The Scripps Research Institute