[gmx-users] regarding range checking error and constraint errors in Lincs algorithm

[Mark Abraham]

shyamala iyer wrote:
> Hi gromacs users,
> 
> I apologize in advance for the long mail, but I have been having some
> trouble with my gromacs simulations lately.
> 
> My system consists of two popypeptide chains and a small ligand (or
> drug like compound). I obtain the required gromacs topologies for the
> ligand/drug from PRODRG dundee server. The steps I use for a MD
> simulation using gromacs is briefly as follows:
> 1) get ligand file with required polar hydrogens and gromacs ligand
> .itp file from prodrg server
> 2) run pdb2gmx using the gmx force field and spce water
> 3) correctly update the protein topology file to include ligand itp
> file header, number of ligand molecules and also make sure the output
> file from pdb2gmx contains the necessary ligand coordinates
> 3) set up box around this system, fill box with solvent (water),
> neutralize system using genion
> 4) set up files to run energy minimization on system, run energy minimization
> 5) set up files to run position restrained dynamics and run position
> restrained dynamics
> 6) setup up files and run Molecular dynamics on the system for ~ 10 ps
> The above set up worked with one of the protease-ligand complex I was
> testing, but had been failing for another viral protease.

This is not a bad start, but check out the advice about NPT and NVT 
equilibration here 
http://wiki.gromacs.org/index.php/Steps_to_Perform_a_Simulation

> The original protease pdb file that I used for the above simulations
> has a small segment missing from chain A. This loop segment was highly
> disorderd in the cystal structure hence, coordinates for this 7
> residue segment is missing in the pdb file. This missing segment is no
> where near where the ligand is bound on the protease. Does a missing
> segment cause problems during siumlations?

The problems you're encountering are consistent with a non-physical 
combination of topology and actual configuration. Depending what you did 
with pdb2gmx, you may find that your topology has a very long bond here, 
or that you have left chemical nonsense here. Either way, the best 
solution will be to add in these residues by hand with some structure 
building software, or to cap these mid-sequence-termini suitably.

> Another general question on the genion function:
> Is there a way I could not have genion be interactive. My systems
> usually have a protein, a ligand and water, and I always choose group
> 13 (SOL) as the selection after running genion. I am asking this
> because there are times when I want to run the same protein with
> several different ligands and I would like a script to run through all
> gromacs steps and not have a particular step be interactive.

http://wiki.gromacs.org/index.php/Making_Commands_Non-Interactive

Mark