[gmx-users] Methodology Check

[Mark Abraham]

On 6/04/2011 3:54 PM, Tom Dupree wrote:
> Greetings all,
>
> I am trying to learn/use GROMACS for evaluating ligand dockings.
>
> After some effort I managed to get a run through EM, equilibration and production with protein and water.
>
> I am now having some difficulty getting a similar run going with a ligand.
>
> I am not ready to talk about the specific problems I am having but I was wondering if someone could check my approach/methodology is correct and if I am doing something GROMACS doesn't like/fundamentally wrong.
>
> I have an .itp from the ATB.

As with any automatically-parametrized species, consider carefully 
whether the result rates to be fit for the job. Various standard advice 
here http://www.gromacs.org/Documentation/How-tos/Parameterization and 
on links from there.

For example, test your ligand topology when isolated in water.

> I take my .pdb and remove the ligand.
> I run PDB2GMX on the no_lig.pdb and get a .top and a new no_lig2.pdb
> I edit the .top
> and add #include "lig.itp"

That needs to go in one of several right places. [moleculetype] sections 
cannot be interrupted (including position restraints sections), and the 
#include must precede [molecules] and follow the forcefield #include.

> I use editconf on the no_lig2.pdb -bt cubic -d 0.5

Probably a bit small for normal interaction lengths of about a 1nm. If 
your protein changes shape a bit, you may find periodic images interacting.

> I use genbox to insert the lig.pdb into the no_lig2.pdb and get complex.pdb
> I use genbox to solvate complex.pdb and return complex.gro as well as an updated .top
> I use grompp to process my em.mdp .top and complex.gro and get complex.tpr<- (my current problem step)
> I use mdrun complex.tpr and get my trajectory
> repeat last two steps for equilibration and production.

Looks generally sound.

> I have not used genion yet, I assumed that it would affect the simulation accuracy but not its ability to run, I am currently +13.

Yes.

> I also realise that I do not have position restraint files for the ligand at the moment so equilibration wont be correct.

genrestr should be useful here.

Mark