RMSD value very high

GROMACS version: 2024.2
GROMACS modification: No

Hi,
I know there is a lot of post related to PBC issues and high RMSD values. I have tried many options, but nothing seems to solve the issue.
I’m getting a surprisingly very high RMSD for my protein-ligand simulation, which looks like this as follows.

When I watched the trajectory it seemed like complex is breaking at several points during the md run.

Plot 1
The recentering steps I have followed,
gmx trjconv -s md_0_10.tpr -f md_0_10.xtc -o md_0_10_center.xtc -center -pbc mol -ur compact

Followed by created a new index group for LIG, and calculated the RMSD with a -nopbc flag which gave a plot like this.

Plot 2
Alternatively, I went through the suggestion from other discussions threads, and performed recentering as followed,
gmx trjconv -f md_0_100.xtc -s md_0_100.tpr -o md_whole.xtc -pbc whole

followed by,
gmx trjconv -s md_0_100.tpr -f md_whole.xtc -o md_cluster.xtc -pbc cluster

And finally,
gmx trjconv -s md_0_100.tpr -f md_cluster.xtc -o md_final.xtc -pbc nojump

which gave a plot as follows (no index group created for LIG),

Additional information:
My ligand is Moracin D, it’s uncharged. It had quite high binding energy when docked in autodock vina.

Question:
Am I missing something in my workflow or applying these commands incorrectly? Any guidance would be greatly appreciated.
Thank you!

One thing I realized.
When I generated the STR file from CGenFF website, it gave a very high penalty. While the guideline says, anything around 10 is fair, and between 10 to 50 needs basic validation, my param penalty was, 136.50 and charge penalty was, 74.184.

Is that why the ligand is behaving weirdly?
Could you please elaborate what type “basic validation” and “extensive validation/optimization” is needed to treat these type of poor analogy?