[gmx-users] First Energy Minimization in a MD

[Justin A. Lemkul]

On 5/12/12 4:25 PM, Lara Bunte wrote:
> Hello
>
> I read the mdp options more than once but I don't understand how to make a good em.mdp file for the first energy minimization, if you plug your molecule in a water box. In my tutorial is this example given.
>
> integrator = steep
> nsteps = 200

This is a relatively small number of steps.  Even fairly standard proteins in 
water often require more iterations in EM.

> nstlist = 10

You should always set nstlist to 1 for EM.

> rlist = 1.0
> coulombtype = pme
> rcoulomb = 1.0
> vdw-type = cut-off
> rvdw = 1.0
> nstenergy = 10
>
>

You do not set any values for emstep or emtol, so you're trusting that the 
defaults are desirable.  The default value for emtol is 10 kJ mol^-1 nm^-1, 
which is often very difficult (if not impossible) to achieve with the steepest 
descents algorithm in single precision.  It is also often unnecessary.

The result of specifying a very low emtol and relatively few steps is that your 
EM is likely to end prematurely, before actually converging.

> 1.) Is it possible to use this as a standard for all first energy minimization.
>

No, for the reasons listed above.  In addition, your settings regarding cutoffs, 
electrostatics methods, etc should be based on what your force field requires. 
Thus, there is no "universal" .mdp file for any simulation process.  In reality, 
during EM the differences will likely be small.  The goal of EM is to produce a 
reasonable starting configuration that can be subsequently equilibrated. 
However, if your .mdp files are wildly different between these two processes, 
you may get unexpected results due to differences in the way the potential (and 
as a result, the forces) is evaluated.

> 2.) Is the choose of parameters here dependent on the force field? I use a CHARMM27, do you think I should change something?
>

Perhaps.  What does your literature reading regarding CHARMM27 tell you are 
appropriate settings?  Particularly important are the cutoffs and vdW method. 
Typically a plain cutoff is not used with CHARMM force fields; a shift function 
is recommended.

> 3.) What are the criteria to choose these parameters? As I said I read the mdp options, so I know what this Option above mean, but I don't know how to know that this are the right parameters?
>
> To say my question in other words: How do I know how the em.mdp file has to look like?
>

The biggest factor is the force field itself and what it requires, as is true 
for all simulation process, EM or otherwise.

-Justin

-- 
========================================

Justin A. Lemkul, Ph.D.
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================