[gmx-users] essential dynamics and flooding

Oskar Berntsson oskar.berntsson at gu.se
Fri Aug 7 10:51:21 CEST 2015

Hi Carsten, and thank you for your help.
>Hi Oskar,
>> On 03 Aug 2015, at 11:27, Oskar Berntsson <oskar.berntsson at gu.se> wrote:
>> Dear all,
>> I am simulating a protein and I want to sample conformations that are not commonly sampled using an equilibrium simulation. As far as I figure, what I want to use is essential dynamics.
>> I am using gromacs VERSION 4.5.5, My protein consists of ca 600 residues. I have run a 250 ns equilibrium simulation and calculated eigenvectors for the ca 600 Calphas using g_covar.
>> I have tested to run essential dynamics using radial acceptence sampling and flooding, but I cannot seem to get it to work as I intend. I have studied the manual and the referenced literature but I cannot find out why I cannot get it to work.
>> 1) When I use radial acceptence sampling the protein rapidly "unfolds". I have tried the commands below to generate my .edi file, in this case with the specific intention to make the protein not unfold.
>> I would expect the simulation to run "maxedsteps" before a new expansion but after the first "eqsteps" the acceptance radius just increases with every step, which quickly deforms the protein and the simulation aborts after ca 2*maxedsteps steps.
>> make_edi -s struct.gro -f eigenvec.trr -eig eigenval.xvg -n index.ndx -radacc 1-3 -slope -1 -maxedsteps 500000 -eqsteps 500000 -mon 1-3 -o sam.edi
>> make_edi -s struct.gro -f eigenvec.trr -eig eigenval.xvg -n index.ndx -radacc 1-3 -slope 0 -maxedsteps 500000 -eqsteps 500000 -mon 1-3 -o sam.edi
>> 2) When I have used flooding I rather have the opposite problem. I cannot force the structure to change one bit over a 50 ns simulation. I have attempted both constant flooding and adaptive flooding.
>> Typically I have generated .edi files as follows
>> make_edi -s struct.gro -f eigenvec.trr -eig eigenval.xvg -n index.ndx -flood 1-3 -alpha 2 -o sam.edi -deltaF0 5 -tau 200
>> make_edi -s struict.gro -f eigenvec.trr -eig eigenval.xvg -n index.ndx -flood 1-3 -tau 0 -Eflnull 10 -alpha 2 -o sam.edi
>> and I have used different deltaF0, tau and Eflnull
>> -deltaF0 = 5-50000000 kJ/mol, tau = 200ps -20 ns and -Eflnull 10-500000000 kJ/mol


>What are the flooding forces you get (e.g. in step 1)?
>Do they at least vary with the values you set for deltaF0 or Eflnull?

Yes, they do increase with increasing Eflnull and deltaF0, but they are typically always ca 0.5-3 at step 100 (the first output step) and Eflnull>100 kJ/mol hardly affects the forces. e.g. Eflnull=10 kJ/mol -> forces 0.63, -1.97, -1.31, Eflnull=500 kJ/mol -> forces 0.84, -2.62, -1,64, Eflnull=50000kJ/mol -> forces 0.84, -2.63, -1.65. For deltaF0 with tau=200ps they incrase from 0 at deltaF0=10 kJ/mol to 0.79, -2.46, -1.54 for deltaF0=50000 kJ/mol. I now tried to decrease the parameter alpha from 2 to 1 and then I get forces 2.68, -8.36 -5.24 at step 100 for Eflnull=50 kJ/mol. I previously tried constant force flooding and forces of ca 10, 10, 10 seems to affect the protein in a desired way.

>  Any help would be greatly appreciated.
>  Best regards
>  Oskar


More information about the gromacs.org_gmx-users mailing list