[gmx-users] position restraint dynamics

chris.neale at utoronto.ca chris.neale at utoronto.ca
Thu Feb 12 19:05:07 CET 2009

If you want to know how much time is required for this step (assuming  
that this step is even required) then that is system dependent and  
something that you would ideally test for yourself. For example, after  
N ps of position restrained dynamics, how does that affect the RMSD of  
the protein from its original structure after 1 ns of unrestrained  
production dynamics -- make the plot and find the asymptote. However,  
I have never seen that done and it is much more common to simply pick  
a time that you guess is "long enough" (sic).

One might run position restrained dynamics in order to equilibrate the  
water (etc) around your protein. If the water is not equilibrated then  
there may be some vaccuum or h-bonding anomaly that may cause the  
protein to dramatically change its local comformation in some region  
during the initial segment of the run. I assume that your position  
restraints are applied in order to avoid this. For some systems, it  
may even be better if you follow your initial position restraints  
simulation by a second simulation in which you restrain only the  
backbone atoms and allow the sidechains to equilibrate -- although  
this may only be necessary if you have some reason to believe that the  
structure that you are starting from is imperfect (perhaps because you  
ignored a bound ligand, etc) and note that side chain rotamerization  
can take many ns.

Best is to read a number of papers and see what people do and how they  
rationalize their decisions. Preferably you would do this from  
simulations conducted with a number of simulation packages (gromacs,  
charmm, amber, etc) as the "equilibration" methods of choice vary  
widely across users of different simulation packages and in order to  
make an informed decision you should be aware of the many different  
options. For example in charmm it is common to slowly turn up the  
temperature from about 100K to the desired temperature over ~10ps, and  
I have never seen this done with gromacs. In gromacs it is more common  
to run a short NVT with frequent velocity reassignment prior to a NPT  
run, although I don't recall seeing any charmm users apply this method.

This is actually an often ignored difference in simulation protocols  
where people have mostly focused on the differing forcefield  
parameters and cutoffs.


-- original message --


we are simulating a protein of 260 a.a length which is a trimer...How does
the mdrun time during the position restraint dynamics affects the Full
Dynamics results???


More information about the gromacs.org_gmx-users mailing list