[gmx-users] confusion about segmentation fault during mdrun

Lan Hua lan.hua at gmail.com
Thu May 20 02:51:34 CEST 2010


Hi Justin,

   I appreciated your quick answers.  So if I understand correctly, using
constraints = hbonds with the time step of 2fs, it should be fine, right?

Thanks,
Lan

On Wed, May 19, 2010 at 3:52 PM, Justin A. Lemkul <jalemkul at vt.edu> wrote:

>
>
> Lan Hua wrote:
>
>> Hi Justin,
>>
>>   Thank you so much for your quick reply and good suggestions. The
>> following is my answer.
>>
>> On Wed, May 19, 2010 at 12:50 PM, Justin A. Lemkul <jalemkul at vt.edu<mailto:
>> jalemkul at vt.edu>> wrote:
>>
>>
>>
>>    Lan Hua wrote:
>>
>>        Hi All,
>>
>>             I understand that the error of segmentation fault may come
>>        from many reasons, but I just couldn't figure out the reason of
>>        this error in my simulations.  I want to run md simulations with
>>        explicit water for 20 structures of one domain (residue 77-148)
>>        of calmodulin (PDB 1CFC).  These 20 starting structures are from
>>        one REMD simulation in implicit water.  The following is what I
>>        did to run simulations for these 20 structures.  I used gromacs
>>        version 3.1.4 with ffamber ports.  The force field is amber03
>>        and water model is TIP3P.
>>
>>
>>    Do you have any particular reason for using software that is eight
>>    years old? You will get a massive performance upgrade with 4.0.7, as
>>    well as the ability to use multiple processors per replica.  In
>>    versions prior to 4.0, you can only use one processor per REMD replica.
>>
>> The reason that I am using gromacs 3.1.4 is to prepare some input files
>> for simulations at folding at home in which version 3.1.4 is recommended.
>>
>>
>
> OK, as long as you've got a reason...
>
>
>
>>
>>          1.  get rid of the steric clash in the starting structure
>>
>>
>>    What do you mean?  Energy minimization?  How did you did do this
>>    prior to step 2 (generating a topology)?
>>
>> I used the "protein preparation wizard" which is implemented in maestro
>> package to do this.   Actually in this wizard, energy minimization is
>> performed on protein.
>>
>>
>>          2.  after doing pdb2gmx, then minimze the protein
>>          3,   use "-bt dodecahedron -d 0.9 -c"  in the command line of
>>        editconf
>>          4,  after doing genbox, first minimize the water with protein
>>        rigid and then minimize the whole system
>>
>>
>>    A lot of these steps are redundant and probably unnecessary.  Some
>> tips:
>>
>>
>> http://www.gromacs.org/Documentation/How-tos/Steps_to_Perform_a_Simulation
>>
>>
>> Thanks for the tips. I went to the link, but I am still a little bit
>> confused about which steps are unnecessary.  You mean step 7 and step 8?  I
>> did this in case simulations at F at H would be crashed.
>>
>>
>
> I just mean the repeated, separate energy minimizations.  I guess there's
> no harm in it, but generally I find that minimizing the protein in vacuo,
> then with and without restraints in solvent, etc. is unnecessary.  I'd
> suggest just building the system (solvent and all), and minimizing the whole
> thing (without restraints).  I don't think you stand to gain anything with
> your procedure.
>
>
>
>>
>>          5,  run md simulation with position restraint for protein
>>        heavy atoms with nose-hoover thermostat for 20ps
>>          6,  run NPT simulations with nose-hoover thermostat and
>>        Parrinello-Rahman thermostat for 500ps
>>          7,  run NVT simulation for another 100ps
>>          8, then energy minimze the whole system again.
>>
>>        Every time, there are always "segmentation fault" in step 6 for
>>        some starting structures which could be different in every try.
>>         I checked the energy, volume, pressure, temperature, etc for
>>        the trajectories which are crashed because of segmentation
>>        fault,  but nothing was wrong.  I roughly checked the trajectory
>>        which looks fine.  I also couldn't find any useful information
>>        from the log file, which looks like the following:
>>
>>
>>    Using weak coupling (i.e. Berendsen) coupling is generally
>>    recommended for initial equilibration.  If a system is far from
>>    equilibrium (as it likely will be after adding patterned blocks of
>>    water with genbox), the N-H thermostat can allow for wild changes in
>>    the temperature of the system, leading to a collapse.
>>
>>    Your temperature coupling groups are also inappropriate:
>>
>>    Tcoupl                   = nose-hoover
>>    tc_grps                  = Protein  SOL  Na
>>    tau_t                    = 0.1      0.1     0.1
>>
>>    Never couple solvent and ions separately; it can lead to instability:
>>
>>    http://www.gromacs.org/Documentation/Terminology/Thermostats
>>
>>
>> These are good suggestions.  Thanks.  So use Berendsen coupling for both
>> temperature and pressure coupling for initial equilibration, for example
>> position restrained NVT followed by NPT, right? I have another
>>
>
> At least for the thermostat, but yes, probably it can't hurt to use weak
> coupling for both temperature and pressure.
>
>
>  question.  If I choose constraints = hbonds instead of constraints =
>> all-bonds in NPT simulation, what will happen?
>>
>>
> You constrain heavy atom-H bonds instead of all bonds.  Using fewer
> constraints may or may not affect the magnitude of the time step you can
> use, but generally X-H bonds are the highest frequency and thus are the
> least stable with long time steps.
>
> -Justin
>
>
>> Best,
>>
>> Lan
>>
>>
>>    -Justin
>>
>>
>>                  Step           Time         Lambda      Annealing
>>                180000      360.00003        0.00000        1.00000
>>
>>          Rel. Constraint Deviation:  Max    between atoms     RMS
>>              Before LINCS         0.045887     47     48   0.004584
>>               After LINCS         0.000020    752    755   0.000003
>>
>>          Energies (kJ/mol)
>>                 Angle    Proper Dih. Ryckaert-Bell.          LJ-14
>>    Coulomb-14
>>           2.08335e+03    1.59908e+02    2.95659e+03    1.17109e+03
>>   1.27711e+04
>>               LJ (SR)  Disper. corr.   Coulomb (SR)   Coulomb (LR)
>>     Potential
>>           4.10779e+04   -1.37728e+03   -2.89916e+05   -5.82443e+04
>>  -2.89318e+05
>>           Kinetic En.   Total Energy    Temperature Pressure (bar)
>>           5.25584e+04   -2.36759e+05    2.96920e+02   -1.07683e+02
>>
>>                  Step           Time         Lambda      Annealing
>>                185000      370.00003        0.00000        1.00000
>>
>>          Rel. Constraint Deviation:  Max    between atoms     RMS
>>              Before LINCS         0.052014     70     71   0.005149
>>               After LINCS         0.000011    214    215   0.000002
>>
>>          Energies (kJ/mol)
>>                 Angle    Proper Dih. Ryckaert-Bell.          LJ-14
>>    Coulomb-14
>>           2.33684e+03    1.42695e+02    2.91169e+03    1.18452e+03
>>   1.28507e+04
>>               LJ (SR)  Disper. corr.   Coulomb (SR)   Coulomb (LR)
>>     Potential
>>           4.06987e+04   -1.37332e+03   -2.88889e+05   -5.83180e+04
>>  -2.88455e+05
>>           Kinetic En.   Total Energy    Temperature
>>
>>        The *.mdp files are also attached.   Any help will be highly
>>        appreciated.  Thank you.
>>
>>
>>        Best,
>>        Lan
>>
>>
>>    --     ========================================
>>
>>    Justin A. Lemkul
>>    Ph.D. Candidate
>>    ICTAS Doctoral Scholar
>>    MILES-IGERT Trainee
>>    Department of Biochemistry
>>    Virginia Tech
>>    Blacksburg, VA
>>    jalemkul[at]vt.edu <http://vt.edu> | (540) 231-9080
>>
>>    http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin
>>
>>    ========================================
>>    --     gmx-users mailing list    gmx-users at gromacs.org
>>    <mailto:gmx-users at gromacs.org>
>>
>>    http://lists.gromacs.org/mailman/listinfo/gmx-users
>>    Please search the archive at http://www.gromacs.org/search before
>>    posting!
>>    Please don't post (un)subscribe requests to the list. Use the www
>>    interface or send it to gmx-users-request at gromacs.org
>>    <mailto:gmx-users-request at gromacs.org>.
>>
>>    Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>>
>>
>>
> --
> ========================================
>
> Justin A. Lemkul
> Ph.D. Candidate
> ICTAS Doctoral Scholar
> MILES-IGERT Trainee
> Department of Biochemistry
> Virginia Tech
> Blacksburg, VA
> jalemkul[at]vt.edu | (540) 231-9080
> http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin
>
> ========================================
> --
> gmx-users mailing list    gmx-users at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/search before posting!
> Please don't post (un)subscribe requests to the list. Use the www interface
> or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20100519/79f8f864/attachment.html>


More information about the gromacs.org_gmx-users mailing list