Molecule going away from protein

GROMACS version: 2023
GROMACS modification: No
Hi, i have performed simulation of protein which has methionine and ATP as ligand. I placed methionine in binding pocket and prepared system using charmm-gui. The visualization of the output gro file has both molecules in pocket, but after npt nvt and final md run, methionine is moving away from protein. Please help me resolve this issue.

did you fix possible periodicity effects? See more here

I tried this, but still methionine is going far from protein.

Have you checked that the interactions are as you’d expect, e.g. that hydrogen bonds etc are present? Have you equilibrated the system with the protein and ligand restrained before you start your production simulations? Are there any water interactions, such as bridging waters, that are important to keep the ligand in place?

The initial PDB structure that i used for system preparation had two interactions between protein and methionine. but after trajectory conversion, neither it has any interactions nor it is giving binding energy in mmpbsa. I used this command for trjconv
echo “1” “0” | gmx_mpi trjconv -s md_0_1.tpr -f md_0_1.xtc -o md.xtc -center -pbc mol -ur compact -dt 1000
Both ATP and Methionine coordinates were taken from other PDB. ATP is in pocket in trajectory but methionine is moving far from protein all the time.
Yes, I equilibrated the system with the protein and ligand restrained before starting production simulations. Please help.

You say that the two interactions are gone after trajectory conversion. Does that mean that they are there before the conversion?

Check if the interactions (and expected water molecules, if any) are there:

  • before and after energy minimization
  • after each equilibration state
  • after the production simulation

Are your forcefield parameters correct? What forcefield are you using? How did you generate the ligand parameters? Is the ligand in the correct protonation state for binding? Are the proton residues in the binding site correctly protonated?

It’s also possible that the simulations (with this force field) do not properly reproduce the bound state or that you have captured the unbinding event (the ligand is not expected to stay forever).