RMSD of ligand calculated by gmx rms doesn't match RMSD from other software

GROMACS version: 2016.1
GROMACS modification: Yes/No

Hi, everyone!
I have a question regarding gmx rms utility. So, I’m interested in calculating the RMSD of a ligand during the course of the simulation.
I use the following command:

gmx rms -s first_frame.pdb -f md_500_united_noPBC.xtc -o rmsd_lig.xvg -tu ns

The trajectory has PBC removed and *pdb file is just the first frame from this trajectory.

Everything works, however, I am quite confused by the results. The final RMSD is around 0.5 nm for the last frame. But when I compare first and the last frames using Pymol and calculate RMSD (first using align for protein structures and then rms_cur to calculate RMSD between ligands) I get value around 2.5 A, which is twice smaller than the value produced by gmx. The RMSD Calculator Plugin from VMD gives the same value as Pymol. Also (visually) the ligand is very stable during the course of simulation.

I wonder if

  • I should try some other options in gmx rms;
  • There is a difference in the alignment algorithms between Pymol and GROMACS (and VMD)?

Another (a bit peculiar) thing I’ve noticed is that the RMSD values are significantly higher when I use *tpr file from trajectory with PBC. But topology file shouldn’t influence the RMSD calculations, or should it?

Thank you in advance for the help!

RMSD is calculated in function of the mass and position of the atoms (see
When you use the option -s pdb_file (or gro_file) you get a warning that say “masses … will be guessed based on residues and atom name …”. This is because pdb and gro file do not have mass information. When you provide -s topol.tpr, you do not get such a warning. The reason is that tpr file contains also information on masses. This can be a source for the difference in RMSD values you observed.

Concerning difference among programs, to properly compare you have to compare the pre-fitting procedure , rmsd formula, and ect.

Best regards