Steered MD doesn't produce expected force graph

User discussions
,
1

GROMACS version: 2022.3
GROMACS modification: No

I am running a pulling simulation on my protein/antibody system but the force graph output is not showing a decrease in force after the protein/antibody have separated. I am pulling the protein away from the antibody with a rate of 0.015 nm/ps with a force of 1000 kJ/mol nm for 500 ps. When I view the simulation in VMD it appears to start separating after 250 ps but in the pullf.xvg file, the force just continues to rise and never stops even after the protein and antibody have separated.

For some background, I started the pulling simulation after 500 ns of a production simulation. I used the prod.gro file which is the last frame of the production, and then minimized it with emtol=1000, ran a short 50ps NPT with all backbone and sidechain restrained, and then ran the pulling simulation. I would have started the pulling simulation immediately but the potential force was too large and crashed the simulation.

Below are the commands I used.

gmx_mpi grompp -f em.mdp -c prod.gro -p topol.top -r prod.gro -o em.tpr -t fit.trr -maxwarn 1
gmx_mpi mdrun -deffnm em

gmx_mpi grompp -f em.mdp -c prod.gro -p topol.top -r prod.gro -o em.tpr -t fit.trr -maxwarn 1
gmx_mpi mdrun -deffnm em

gmx_mpi grompp -f pull.mdp -c npt.gro -p topol.top -o pull.tpr -r npt.gro -n index1.ndx -t npt.cpt -maxwarn 1
gmx_mpi mdrun -deffnm pull -pf pullf.xvg -px pullx.xvg

Has anyone encountered this problem or have any ideas on how to fix it?

2

Your pulling speed is very high: 15 nm per ns. Thus you will measure viscous friction of the molecules moves through the solvent.

The proper solution is to not use steered MD, but umbrella sampling or an enhanced sampling method like AWH. But in practice it is often impossible to get reasonably converged results as the sampling is often too challenging.