[gmx-users] Atom names mismatch

[Mark Abraham]

George Abadir wrote:
> Thanks for your reply. I built the molecule using PYMOL, then used 
> pdb2gmx to generate the topology. It seems that pdb2gmx reorders the 
> atoms. Then in the simulation I used the initial PDB file generated by 
> PYMOL. 

Right. Don't do this. For input to grompp and/or other GROMACS tools, 
use the structure that pdb2gmx writes out, since that is now consistent 
with the re-ordering in the topology. This will immediately solve the 
issue for the future. In future, generate your topologies interactively, 
and check for warnings, etc before running simulations.

You can partly validate, or possibly totally invalidate the old 
simulations by doing a quick repeat calculation with a correct match of 
topology and starting structure. If valid, the logfile energy numbers 
should remain identical for several picoseconds. If not valid, they'll 
diverge in that time.

> Does this make things any more reasonable. Because actually, this 
> happened in five different simulations till now!!!! So, I don't think it 
> is a mere coincidence. The molecule at the end is intact and all the 
> atoms are in their respective positions (i.e., the carbon was not 
> actually exchanged with hydrogen as it should be from the warning). It 
> would very helpful if someone could see how these simulations continued 
> without deforming the molecule.

Mark