Question about COARSE GRAINED IN LAMMPS

Hi lammps users,
Could i use MARTINI FORCE FIELD in LAMMPS?

Hello, Nicola:

Unless the MARTINI model has been significantly revised in its recent incarnations, it is not easy to implement as written in LAMMPS. The fundamental issue is that the MARTINI model has treated the backbone as if it were a single “frozen” entity, which for reasons of parallel communication efficiency have not been allowed in LAMMPS. (Essentially, the entire backbone would need to be on a single processor to function correctly.)

So, if this constraint has been relaxed, then it should at least be possible to run “as is.” However, if not, then substantial coding would be required to complete the implementation.

—AEI

Forgot to include this in my last email—you should see the following discussion from the mailing list for more information:

http://lammps.sandia.gov/threads/msg17360.html

—AEI

I think Nicola did not suggest to model rigid polymers, so for most cases the answer is yes.

Francis

Forgot to include this in my last email—you should see the following discussion from the mailing list for more information:

http://lammps.sandia.gov/threads/msg17360.html

—AEI

Forgot to include this in my last email—you should see the following
discussion from the mailing list for more information:
http://lammps.sandia.gov/threads/msg17360.html
—AEI

I think Nicola did not suggest to model rigid polymers, so for most cases
the answer is yes.

Thanks Ahmed and Francis

    Thanks for bringing up the SHAKE issue. As Steve and Francis
suggested, you can probably get around SHAKE limitations using
stiff-enough bonds and bond-angles, and dihedral forces. I don't know
how this would effect the timestep, or whether LAMMPS would be as fast
as NAMD or Gromacs.

   Molecule-builder software:

    It looks like there are some pretty nice software tools already
available to build these systems for NAMD (and LAMMPS). I have not
tried this, but apparently, you can make LAMMPS DATA files for these
martini models using the "cgtools" and "topotools" plugins for VMD,
which are available here:
http://www.ks.uiuc.edu/Research/vmd/plugins/cgtools/
http://sites.google.com/site/akohlmey/software/topotools
    Use "cgtools" to build the system in NAMD format, and topotools to
create the LAMMPS DATA file containing coordinates and topology
information. You would still need to create a list of pair_coeff,
bond_coeff, angle_coeff, dihedral_coeff, and improper_coeff commands
containing all the force-field parameters for the molecules in your
system. I'm guessing these parameters can probably be extracted from
the .PAR files created by "cgtools".

   At some point, I planned to create Martini versions of water,
lipids, cholesterol, and nucleic and amino-acids in moltemplate format
and post them. (I never pass up an opportunity to name-drop
"moltemplate".) It's a bit of work, but if there is significant
demand for this, I might do it, eventually.

Andrew

P.S. There is an "ltemplify.py" program which can convert individual
molecules in LAMMPS (DATA+IN) format into moltemplate (LT) format. If
somebody actually has a full set of amino-acid martini-models in
LAMMPS format, please email them to me. I'd love to convert them and
post them (with credit to the original author, of course).

Dear Andrew:

Yes, using stiff bonds requires much smaller time step. In my experience, coarse-grained benzene with large force constant is unstable using step above 5 fs.

For a moderate-sized surfactant system (including LJ and PME), LAMMPS is not faster than GROMACS. However, LAMMPS has more functionalities and is growing fast.

It should be helpful to develop such a CG builder, so thank you for your efforts.

Francis