I encountered an issue while attempting to run GROMACS on a protein that includes an ATP molecule. The GROMACS input files were generated using CHARMM-GUI, but when I attempted to run the simulation, I encountered an error. To resolve this, I tried minimizing the structure using Chimera, but the issue persisted. Additionally, I used Galaxy Refine for refinement; however, this approach led to the removal of the ATP molecule from the refined models.
From what I can see, the energy minimization stops because it gets infinite forces, and then when you go on and start the NVT simulation the run crashes instantly for the same reason.
I suggest you to take a look at where the energy minimization is failing. The slurm log file says that the atom where it is having problems is the 14527. Check the structure with some molecule visualization software (VMD, pymol, etc.) and try to understand why this is happening. It may be sufficient to move bit away the atom and/or modify the energy .mdp file parameters to make the equilibration more gentle, but I doubt so. If this was directly the output of CHARMM-GUI, re-building the box might be the safest and fastest approach. Sometimes the webserver fails in putting the atom in the box and returns overlapping structures!
