[gmx-users] freezing problem

[Oleg V. Stroganov]

Hello , Gromacs users

I am using Gromacs to study protein-ligand interactions. First, I
tried to run MD with a simple system: frozen protein and one molecule
of a ligand. I have encountered strange result: the ligand has moved a
liitle bit towards protein surface and then got almost frozen. It did
not look as a binding process in common chemical sense: all atoms of
the ligand stopped moving/oscillating, only one CH3-group continued
rotating as crazy. Velocity distribution of moving atoms was far far
from the maxwellian. The temperature was 800K.

Another exercise was MD simulation of the same frozen protein, ligand and
water layer around. I have filled with water only the active site
cavity of the protein, and used the outer frozen layer of water as a
wall to prevent evaporation of the inner solvent. During the
simulations (20 ns at  300K) 'mobile', i.e. non-frozen water molecules
did not diffuse within the allowed volume, they only oscillated
instead. The same was with the ligand.

These results, the absense of convection, however had strange
reproducibility: identical runs (with different seeds) showed
different results. One simulation has got frozen, while another -
looked normal. Some other simulations have demonstrated freezing from
time to time during the trajectory.

Formally, temperature (from logs and edr) was good.

Could anybody comment on these strange issues?

-- 
Best regards,
 Oleg Stroganov,
 Moscow State University student
 mailto:Oleg_Stroganov at newmail.ru