[MMTK] some general questions

hinsen at cnrs-orleans.fr hinsen at cnrs-orleans.fr
Fri Apr 23 18:08:13 CEST 2004

On 22.04.2004, at 16:00, Marc Gasser wrote:

> 1. 	I wanna simulate a protein-ligand complex by MMTK,
> 	what kind of charges do you suggest to be assigned
> 	to the ligand?
> 	what's wrong with Gasteiger charges on a protein?

That's not the relevant question. The partial charges of a force field  
are not physical observables anyway, they are parameters for the force  
field. They should therefore be determined as prescribed by the force  
field. I am not aware of any force field that is suitable for proteins  
and uses Gasteiger charges, but one could perhaps construct one. Amber  
(the force field in MMTK) is based on charges obtained from fitting to  
quantum chemical calculations in a well-defined way (as described in  
the accompanying paper).

> 2.	how can I add counterions to a universe? is there a
> 	possibility to play with different pH values and different
> 	protonation states of the protein or ligand? what about
> 	tautomerism?

Counterions are just charged atoms, so just create an atom and change  
its charge, or create a new atom type with the right charge. There is  
no code in MMTK that figures out for you how many counterions you  
should add, so you have to do that yourself in the application script.

Note that explicit counterions are not necessarily the best approach.  
In a typical biochemical simulation, the concentration of the  
macromolecules is usually much too high, and then the concentration of  
counterions is much too high as well. This can create detrimental  
artefacts. It may be better to use Ewald summation for calculating the  
electrostatic interactions and rely on its built-in charge  
neutralization (which is physically equivalent to adding a homogeneous  
neutralizing charge density to the system). But that depends on your  
application of course.

DIfferent protonation states are considered different residues by MMTK.  
When creating a protein, you must specify which ones you want. Again  
there is no code that figures out for you which protonation state is  
appropriate at a given pH.

> 3. 	Is it possible to set positional restraints on a set of
> 	atoms, and to reduce the restraints during the simulation,

Yes, check MMTK.ForceFields.Restraints. The restraints become a part of  
the force field, so if you want to add, remove, or modify restraints,  
you just assign a new force field to the universe.

> 	the fixed attribute of atoms is freezing them totally, if I
> 	understood correctly, isn't it?


> 4.	What about modeling one or more mutations within a proteinchain
> 	of a PDB-file, is there a method which is generating coordinates
> 	for the new aminoacids?

No. It's rather easy to do unless the new sidechain is much longer than  
the old one. You can look at the code that replaces cystine by cysteine  
as an example (in MMTK.Proteins).

> 5.	somehow I didn't really understand the term 'configuration' in the
> 	manual. Is there a difference between 'conformation of a protein'
> 	and 'configuration of a protein', or is 'configuration'
> 	all information and 'conformation' information of positions.

In the context of simulations, the two terms are synonyms.  
Experimentalists often use "conformation" for what a theoretician would  
call a local minimun of the free energy function, i.e. an energetically  
stable configuration.

> 6.	this is from a example from the MMTK distribution.
> 		minimizer = SteepestDescentMinimizer(world, step_size =  
> 0.05*Units.Ang)
> 		minimizer(steps = 100)
> 	what exactly does the second line? I mean it starts the minimization
> 	but why not something like
> 		minimizer.minimize(steps=100)

Because that's the way the code was written. Several other choices  
would have been reasonable as well, of course. The view taken in MMTK  
is that a minimizer is a kind of function, so it is called like a  

Konrad Hinsen
Laboratoire Leon Brillouin, CEA Saclay,
91191 Gif-sur-Yvette Cedex, France
Tel.: +33-1 69 08 79 25
Fax: +33-1 69 08 82 61
E-Mail: hinsen at llb.saclay.cea.fr

More information about the mmtk mailing list