[gmx-users] charmm and oplsaa simulation results don't match

[Chris Neale]

water is part of the ff, therefore see previous comments of Berk and myself. For a reference, see Ponder and Case 2003 
(http://www.ncbi.nlm.nih.gov/pubmed/14631816), and while one of the pmfs that they present is quantitatively incorrect, the
conclusions are valid: the different protein/water ff's have not entirely converged over time.

-- original message --

hi,
  
  Thanks for the suggestions. Actually the microstates are revisited in the 100ns span and one of them is occupied for longer time span than the others.
I had broken the 100ns trajectory(oplsaa) in 3 parts and took starting structures from each of them and run simulations and obtained compatible results. However what's disturbing is that in the charmm run, the configuration of the staring structures changes within the first few picoseconds and then remains in the new configuration for the rest of the simulation time.It means that the starting structure changes into another structure and retains that for the rest of the simulation and that happens to be the most occupied state in the oplsaa run.
I may be wrong, but I think somewhere i am having problem in the way charmm and tip3p water model works ie. it may be that the L-J potentials of the hydrogen atoms are not treated properly.
Surprisingly, I had run charmm with spc water model(by mistake), and although the most occupied state remains the same but the microstates are obtained.

Thanks in advance
Sarbani