[gmx-users] Urea molecule deformed in urea-water simulations using ffamber99

[Mark Abraham]

George Abadir wrote:
> Hi,
>     I am doing a simple simulation of one urea molecule dissolved in 
> water. I am using the ffamber_urea.itp introduced in the AMBER port for 
> GROMACS. I build a box using editconf and solvate the urea molecule 
> using genbox with the fftip3p.itp water model also included with the 
> AMBER port. After energy minimization (which converges to machine 
> precision but not to the desired emtol=100. Actually the maximum force 
> is about 550 for a cubic box), I find that there is one extra bond in 
> the urea molecule between the oxygen atom and the nitrogen atom N1 and 
> this bond was not there prior to energy minimization.

No, there's no "extra bond". Your visualization software probably uses 
some heuristic to decide whether to *render* a bond between two atoms... 
it's certainly not using your .top file where you defined your bonds, or 
the .tpr generated from it, since neither of these will now have an 
extra bond. Does the structure actually look bad? Do the contents of 
your .top file (and included .itp files) actually mean what you think 
they mean?

> I gave it a shot with the production simulation after that, but sure 
> enough, it gave a segmentation fault (I am not surprised with that 
> because of the bad energy minimization).
> Can anyone help me know why the energy minimization fails and the urea 
> molecule becomes deformed? Your help is much appreciated.

Read the output to pdb2gmx, editconf and genbox carefully.

Mark