Further to my last message we have managed to use a tcl script in VMD to calculate the order parameters for the POPC chains in the CHARMM36 simulation. I have put the values for this calculation and the values I obtained using g_order (over the same period in the same trajectory) below (this is for the sn-2 chain, for the sn-1 chains then the 2 methods produce very similar results). I think this shows that there is a problem with g_order and the calculation of the order parameter parameters for the first carbon in the double bond of the unsaturated tail.

           carbon       g_order         VMD script

           2            0.0967924       0.095035
           3            0.208143        0.207425
           4            0.203251        0.203339
           5            0.214236        0.213071
           6            0.192448        0.191573
           7            0.18472         0.182894
           8            0.101735        0.108403
           9            0.134363        0.061669        ***
          10            0.0444899       0.049259
          11            0.0703591       0.077746
          12            0.116684        0.115886
          13            0.122237        0.121924
          14            0.126424        0.125857
          15            0.114384        0.114213
          16            0.103868        0.103317
          17            0.0806387       0.080544

If nobody can see that I have done anything obviously wrong then I will file an issue regarding this on redmine.



Thomas Piggot wrote:

Both the trajectories are 100 ns in length (plus I have a repeat of each simulation using different starting velocities). The results from a block analysis, and analysis of the repeat simulations, give almost identical results to those in my previous message. The other thing is that the remainder of the order parameters also give sensible numbers, both for the saturated carbons in the rest of the oleoyl chain and in the palmitoyl chain.

Regarding the mdp settings, then I used the same parameters as given in the papers from which I quoted the previous values. For CHARMM36, this includes a 1 fs timestep, the CHARMM TIP3P water (tips3p), constraints = hbonds and a switching of the vdW interactions between 0.8-1.2 A. For the GROMOS53A6L simulations then this is with a cut-off for the electrostatic interactions of 1.4 nm and a reaction-field correction applied beyond the cut-off, a vdW cut-off of 1.4 nm, rlist of 0.8, nstlist every 5 steps, a 2 fs time step and the SPC water model. I can provide full mdp's if necessary. Analysis of other membrane properties from the simulations (such as area and volume per lipid, area compressibility, electron density) all give comparable results to those previously published for these two force fields.

All of this leads me to suspect that there may be a problem in the way g_order calculates the order parameter for the unsaturated carbons, however I am by no means certain of this. Currently we are trying to calculate the order parameters from the all-atom CHARMM36 simulations using a different bit of code, I will report back with the results if we can get this to work.



Justin A. Lemkul wrote:
Thomas Piggot wrote:
Hi Igor,

Thanks for the reply but I think you slightly missed the point I was trying to make. I followed the approach you mention for the double bond and (with the two force fields I tried) got the values I discussed in my last email, which when compared to the published values seem to be incorrect.

Just to confirm, my index file (for the unsaturated calculation) contains (below is just pasted from make_ndx):


  0 C28                 :    72 atoms
  1 C29                 :    72 atoms
  2 C210                :    72 atoms
  3 C211                :    72 atoms

and for GROMOS53A6L:

  0 C1H                 :   128 atoms
  1 C1I                 :   128 atoms
  2 C1J                 :   128 atoms
  3 C1K                 :   128 atoms

My command for g_order (this is just for CHARMM36 but I use the same command for GROMOS53A6L on different files) is:

g_order_4.5.4 -s for-md-popc-charmm36.tpr -f md-popc-charmm36.xtc -n chain2_unsat.ndx -od deut_chain2_unsat.xvg -unsat

If anyone can see what might be going wrong then any help would be greatly appreciated.

Are your results converged? From the g_order command, you're considering the whole trajectory. I'd suggest the usual block averaging approach to see if you're converged.

There's nothing wrong with the way you're calculating the order parameters, so the only other thing I would suspect (other than lack of convergence) is that the .mdp settings are somehow giving unexpected results. Membrane properties can be very sensitive to vdW and electrostatics methods.


Dr Thomas Piggot
University of Southampton, UK.
gmx-users mailing list    gmx-users@gromacs.org
Please search the archive at 
http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org.
Can't post? Read http://www.gromacs.org/Support/Mailing_Lists

Reply via email to